密码与魔法

Codes and Magic

这个神话可能和语言本身一样古老。世界上存在着咒语：可以通过程序性话语的力量改变现实的咒语。婚姻誓言、法庭判决、萨满诅咒：这些词语都是改变现实的代码。这是一个古老而诱人的想法。1^从《创世纪》中的逻各斯到众多确定上帝“真名”的宗教传统，人类一直坚信某些咒语不仅仅是描述世界，更是创造世界。这又有何不可？语言始终游走于现实与对现实的描述之间那条难以逾越的界限。一个想法越结构化、抽象化、深奥难懂，我们就越不可能在不先知道一个名字的情况下洞察其本质。

The myth is probably as old as language itself. There are spells in the world: incantations that can transform reality through the power of procedural utterances. The marriage vow, the courtroom sentence, the shaman’s curse: these words are codes that change reality. It is an old and attractive idea.¹ From the logos of Genesis to the many religious traditions identifying the “true names” of God, humanity has persistently believed that certain invocations do not merely describe the world but make it. And why not? Language has always operated at the troubled boundary between reality and the description of reality. The more structured, abstract, and esoteric an idea, the less likely we are to divine its substance without first gleaning a name to call it by.

如今，我们的语言横跨众多领域：程序化计算机语言、电影和新媒体的批判性语言、克里奥尔语、虚构语言、新话、表情符号。在我们的认知中，每一种语言都赋予符号和意义某种神奇的力量；每一种语言都基于现实与表象之间内在的张力，孕育出文化力量。口语与抽象符号系统（尤其是数学）之间的联系，为数字、普遍真理和现实基本结构之间的神秘联系开辟了新的途径。犹太教卡巴拉教派、艾萨克·牛顿对炼金术的痴迷，以及像黄金分割率这样的数学图形的生物学例证，都强化了一种特定的形而上学观念，即宇宙的底层存在着某种逻辑顺序、某种语法和符号词汇。

Today our languages sprawl across many registers: procedural computer languages, critical languages of film and new media, creoles, fictional languages, newspeak, emoji. In our perception, each of those registers ascribes certain magical powers to symbols and meaning; each of them generates cultural power based on the inherent tension between reality and representation. The link between spoken language and abstract symbolic systems, particularly mathematics, has created new avenues for mystical connections between numbers, universal truths, and the fundamental structure of reality. Jewish kabbalah, Isaac Newton’s fascination with alchemy, and biological examples of mathematical figures like the Golden Ratio all reinforce a particular metaphysical notion that some logical order, some grammar and symbolic vocabulary, underlies the universe.

在对这些问题的争论中，哲学家和数学家对符号语言的理解日益深奥，为当代计算时代奠定了基础。从集合论和符号逻辑的内核，到数据驱动机器学习的最新表述，计算投射出一种文化阴影，这种阴影源于这种悠久的魔幻思维传统。随着计算几乎改变着文化生活的方方面面，我们讲述的关于它的故事，以及神话与理性之间的平衡，将在决定我们能够认知和思考什么方面发挥重要作用。语言在世界上拥有力量，并且在某种程度上可以定义世界。符号逻辑一旦被运用，就能对现实产生程序性的改变。

In debating these questions, philosophers and mathematicians developed increasingly sophisticated understandings of symbolic languages, laying the groundwork for the contemporary era of computation. From its bones in set theory and symbolic logic to the latest articulations of data-driven machine learning, computation casts a cultural shadow that is informed by this long tradition of magical thinking. As computation transforms almost every aspect of cultural life, the stories we tell about it, the balance of myth and reason, will play a major role in determining what we can know and think. Language has power in the world, and may in some sense define the world. When enacted, symbolic logic can effect procedural alterations to reality.

这里的关键词是“制定”。本书揭示了算法——这一不起眼的计算载体——的根源不仅在于数理逻辑，还在于控制论、意识和符号语言的魔力等哲学传统。为了理解算法，我们需要揭示这些根源，然后构建一个新的“算法阅读”模型，该模型融合了对抽象和过程的深刻理解。算法将理想化的计算空间中的概念部署到混乱的现实中，并将其应用于我称之为“文化机器”的东西：抽象、过程和人的复杂组合。算法将理论理念付诸实践，在具体实施的细节上，两者之间总会存在差距。这种实施差距是我们最需要了解的，也是我们最常误解的关于算法系统的事情。理解我们如何能够了解这一点，需要人文学科的批判性方法。这就是算法阅读：一种应对计算固有复杂性以及这种复杂性与人类文化交织时产生的模糊性的方法。

The key term here is “enacted.” This book uncovers how the humble vehicle of computation, the algorithm, has its roots not only in mathematical logic but in the philosophical traditions of cybernetics, consciousness, and the magic of symbolic language. To understand the algorithm we need to uncover those roots and then build a new model of “algorithmic reading” that incorporates a deep understanding of abstraction and process. The algorithm deploys concepts from the idealized space of computation in messy reality, implementing them in what I call “culture machines”: complex assemblages of abstractions, processes, and people. Algorithms enact theoretical ideas in pragmatic instructions, always leaving a gap between the two in the details of implementation. The implementation gap is the most important thing we need to know, and the thing we most frequently misunderstand, about algorithmic systems. Understanding how we can know that requires the critical methods of the humanities. This is algorithmic reading: a way to contend with both the inherent complexity of computation and the ambiguity that ensues when that complexity intersects with human culture.

挖掘

Excavation

上述题词中，尼尔·斯蒂芬森（Neal Stephenson）的开创性赛博朋克小说《雪崩》（Snow Crash）中的“nam-shub”（南舒布）是古老的苏美尔咒语，其魔力感染了现代硅和二进制逻辑的基底。一位狂妄自大的亿万富翁开始挖掘刻有真实咒语的苏美尔泥板，这些咒语曾经拥有直接操控人类思维的力量。苏美尔人听到“nam-shub”后，会将其视为程序性语言，即一组能够改变其思维和世界的指令。

In the epigraph above, the nam-shubs of Neal Stephenson’s seminal cyberpunk novel Snow Crash are ancient Sumerian incantations whose magic infects the modern substrates of silicon and binary logic. A megalomaniacal billionaire begins digging up Sumerian clay tablets inscribed with actual spells that once had the power to directly program human minds. A Sumerian hearing a nam-shub would ingest it as procedural language, a set of instructions that could alter her mind and her world.

斯蒂芬森将苏美尔神话的魔法传统改编成近未来的数字文化，虚拟现实和无处不在的计算都已十分成熟。小说将 nam-shub 想象成一种语言算法或黑客手段，将神灵恩基的诗意咒语变成了一套人类思维的操作指令。斯蒂芬森将这段晦涩的神话历史放在黑客手中，使读者更加熟悉，而到 1992 年这本书出版时，黑客已经沉浸在他们自己的当代神话中。在《雪崩》中，主角 Hiro 讽刺性地定义了黑客形象的标志，黑客工作在庞大的文化系统边缘，通过技术技能、理想主义动机和对传统习俗的漠视进行关键干预。Hiro 是一个完全符合科技记者史蒂芬·利维在《黑客》中所描述的骗子原型的人物；这个角色诞生于 1984 年硅谷先驱斯图尔特·布兰德的黑客大会上^。2小说中的计算系统，从各种安全系统到元宇宙本身，都是由黑客创建的，并受到他们的操纵。

Stephenson adapts the magical tradition of Sumerian myth to a near-future digital culture where virtual reality and ubiquitous computation are well established. The novel imagines the nam-shub to be a linguistic algorithm or hack, turning the poetic spells of the god Enki into a set of operating instructions for the human mind. Stephenson makes this rather obscure mythic history more familiar to his readers by putting it in the hands of hackers, who by the book’s publication in 1992 were already imbued with their their own contemporary mythos. In Snow Crash, Hiro Protagonist satirically defines the icon of the hacker figure, working at the periphery of monolithic cultural systems to make crucial interventions through technical skill, idealistic motivation, and a blithe disregard for traditional mores. Hiro is a character right out of the trickster archetype that technology journalist Steven Levy chronicles in Hackers; a character who came to life around Silicon Valley pioneer Stewart Brand’s Hackers Conference in 1984.² The computational systems of the novel, from the various security systems to the Metaverse itself, were created by hackers and are subject to their manipulations.

作为赛博朋克流派的巅峰之作，《雪崩》将黑客美化并奉为神圣，他们是计算现实中强大而反复无常的构建者。黑客是骗子和叛逆者，他们施展的技术威力堪比魔术（在《黑客》、《潜行者》和《黑客帝国》等电影中，他们经常被描述为近乎魔法，在这些电影中，源代码变成了媒体评论家全温迪·惠京所说的“源码术”）。3^他们对代码的控制力如同萨满教，他们作为边缘人物的角色强化了这样一种观念，即计算本身容易受到神秘的外部力量的影响。这个故事现在对我们来说已经非常熟悉，利维、布兰德、全温迪、数字文化学者弗雷德·特纳和其他许多人都对其进行了详细的记录^。4

As a high-water mark in the cyberpunk genre, Snow Crash both embellished and consecrated hackers as potent and capricious architects of computational reality. Tricksters and rebels, hackers performed feats of technological prowess akin to magic (and often depicted as quasi-magical in films like Hackers, Sneakers, and The Matrix, where the source code becomes what media critic Wendy Hui Kyong Chun calls “sourcery”).³ Their power over code was shamanic, and their very roles as peripheral figures reinforced the notion that computation itself was susceptible to mysterious external forces. This story is by now deeply familiar to us and well chronicled by Levy, Brand, Chun, digital culture scholar Fred Turner, and many others.⁴

然而，正如《雪崩》题词所暗示的那样，黑客与其说是这部小说的主体，不如说是情节传播的媒介，它是一种源自古老纳姆舒布（nam-shub）的模因病毒。在黑客聚光灯下的身影背后，是他或她汲取力量的更为阴暗的领域：计算空间本身，包括源代码、程序语言、黑客手段和巧妙的数字技巧。随着计算机变得越来越普及和易于访问，通过我们不断增长的屏幕、键盘输入和低声的询问，计算空间开始变得更加可思考和可触及。我们不仅从逻辑上越来越接近计算空间；我们现在与它已经亲密无间，分享我们最私密的记忆，并在从爱情到房地产等各种事务上接受算法的指导。

But, as the Snow Crash epigraph suggests, hackers are not so much the subjects of this novel as the medium through which the plot propagates, in the form of a memetic virus derived from an ancient nam-shub. Behind the spotlit figure of the hacker is the gloomier region from which he or she draws power: the space of computation itself, of source code, procedural language, hacks, and clever digital tricks. As computers become more ubiquitous and accessible, that space of computation has started to become more thinkable and tangible through our proliferating screens, keystrokes, and muttered queries. We are not just getting logistically closer to that space of computation; we are now intimate with it, sharing our most personal memories and accepting algorithmic guidance on matters ranging from love to real estate.

《雪崩》将这个故事延伸为一个关于“计算即魔法”的当代寓言。那些将这种病毒带到这个世界的黑客之所以这样做，是因为他们特别容易受到一种神经语言学感染。在《雪崩》中，计算空间通过一种代码——一种在意识和世界中运作的程序性语言——触动你。

Snow Crash extends this story into a kind of contemporary parable for computation-as-magic. The hackers who carry this virus into the world do so because they are particularly vulnerable to a kind of neurolinguistic infection. In Snow Crash, the space of computation touches you through a kind of code, a procedural language that operates in consciousness as well as in the world.

正因为语言作为一种智力技术占据着特殊的地位，它作为认识论层面或媒介的作用一直受到从柏拉图到约翰·塞尔等哲学家们的关注。在代码的语境下，语言可以重塑世界和思维。如果没有合适的词汇，表达某些想法会非常困难，有时甚至几乎不可能，尤其是在技术系统的语境下。例如，人类语言似乎会以可预测的顺序习得不同的颜色词汇，而这些词汇的存在，出于文化目的，使得颜色的现实性成为可能：如果没有“绿色”这个词，光谱中的这一波段就会消失在邻近的“蓝色”概念中。5^这些语言的使用者和我们其他人一样，拥有相同的眼睛和相同的生物视觉器官，但他们的大脑中却没有“深层结构”和“神经语言通路”来解析视觉光谱中这一特定的波段。在某种程度上，我们感觉中枢的普遍生物功能让位于语言构建和优先化体验的相对力量。

Precisely because language occupies a special status as an intellectual technology, its role as an epistemological layer or medium has interested philosophers from Plato to John Searle. In the context of code, language can reformat the world and the mind. Without the right vocabulary, it is difficult, sometimes almost impossible to articulate certain ideas, particularly in the context of technical systems. Human languages seem to acquire different color words in a predictable sequence, for example, and the existence of the word makes possible, for cultural purposes, the reality of the color: without the word for “green,” that band of the spectrum disappears into the neighboring concept of “blue.”⁵ Speakers of these languages have the same eyes, the same biological apparatus of vision as the rest of us, but they do not have the “deep structures,” the “neurolinguistic pathways” in their brains parsing out this particular band of the visual spectrum. At some point the universal biological equipment of our sensorium gives way to the relative power of language to structure and prioritize experience.

语言这种独特力量的必然结果是，有些咒语是无法被忘记的：当我们诠释和理解它们时，它们会永久地改变我们。在《雪崩》中，斯蒂芬森设想，神恩基故意释放了一种名为“nam-shub”的病毒，摧毁了苏美尔语这种通用的传输系统。这种病毒“像蛇一样缠绕在人类脑干上”，扰乱了人类理解其他苏美尔信号的能力。6 ^《雪崩》将这一时刻与神话中的巴别塔联系起来，使“nam-shub”成为一种曾经失传但现在又重新获得（并被用于邪恶目的）的通用语言的遗物。因此，斯蒂芬森挖掘了语言作为咒语的更深层神话。如果代码可以具有魔力，而黑客是它的巫师，那么我们仍然将其视为在认知与现实交汇处运作的符号系统。通过赋予代码图形以文化力量，我们也认可了它在平台上运行的概念：人类可能运行通用操作系统的想法。

The corollary to this distinguishing power of language is that some incantations cannot be unheard: they permanently alter us when we interpret and understand them. In Snow Crash Stephenson imagines that the god Enki deliberately destroyed the universal system of transmission that was the Sumerian language by releasing a nam-shub virus that, “coiled like a serpent around the human brainstem,” scrambled humanity’s ability to understand other Sumerian signals.⁶ Linking this moment to the mythic Tower of Babel, Snow Crash makes the nam-shub a relic of a universal language once lost but now regained (and being put to nefarious use). Thus Stephenson taps into the much deeper mythos of language as incantation. If code can be magical and hackers are its shamans, we still recognize it as a symbolic system that operates at the intersection of cognition and reality. By investing the figure of code with cultural power, we also endorse the notion that it functions on a platform: the idea that humanity might run a universal operating system.

所以代码可以是神奇的，代码可以改变世界，代码可以改变思想。但这实际上是如何运作的？在计算空间中，实体和操作结构是什么？在《雪崩》的新苏美尔操作系统中，有“我”，即体现重要文明概念的特定语言单位。这种比喻在其他传统中也很常见，像普罗米修斯或郊狼这样的骗子人物从众神那里窃取了概念技术（例如火）。从某种意义上说，“我”是可以在人群中运输和以某种方式部署的物体。但它们也是知识体系、规则和程序集，可以在实践中实施。它们是独立于人类从业者而存在的技术实体，但通过文化媒介运作。它们是算法。

So code can be magical, code can change the world, and code can change the mind. But how does this actually work? What are the entities, the structures of operation in that space of computation? In Snow Crash’s neo-Sumerian operating system there are me, particular units of language that embody vital civilizational concepts. This trope is familiar from other traditions as well, where trickster figures like Prometheus or Coyote steal conceptual technologies (e.g., fire) from the gods. In one sense me are objects that can be transported and somehow deployed among populations. But they are also bodies of knowledge, sets of rules and procedures, that can be implemented in practice. They are technical entities that have their own existence independent of their human practitioners, but which operate through the medium of culture. They are algorithms.

本书探讨的是算法作为计算的载体或工具：它是计算空间、文化系统和人类认知交汇的对象。我们需要更深入地理解算法，才能理解计算系统如何改变当今世界。从这个意义上讲，这是一项素养练习，一项开发“算法阅读”世界的实验。人文学科和批判性阅读（算法阅读）方法论的作用对于有效应对计算与文化交汇处存在的模糊性和复杂性至关重要。但本书也是对一个理念从其当代文化存在到其哲学基础的性格研究。《雪崩》巧妙地展现了算法将计算、神话和文化空间连接在一起时所产生的张力。与其说这是一个关于代码力量的故事，不如说是一个关于代码如何在算法的幌子下对现实、理想和想象进行尴尬的拥抱的故事。

This is a book about the algorithm as the vehicle or tool of computation: the object at the intersection of computational space, cultural systems, and human cognition. We need a deeper understanding of the algorithm in order to understand how computational systems are transforming our world today. In that sense this is a literacy exercise, an experiment in developing an “algorithmic reading” of the world. The role of the humanities and methodologies of critical reading—algorithmic reading—are vital to effectively contend with the ambiguity and complexity at play in the awkward intersection of computation and culture. But this is also a character study of an idea from its contemporary cultural presence to its philosophical foundations. Snow Crash neatly illustrates the tensions at play when algorithms stitch together computational, mythic, and cultural spaces. It’s not so much a story about the power of code but its awkward embrace of the real, the ideal, and the imaginary in the guise of the algorithm.

算法作为一种近乎神秘的实施知识结构，这种形象既普遍存在，又鲜为人知。如今，我们从未像现在这样接近实现完全实施的计算知识这一隐喻，因为平台和系统的爆炸式增长正在重塑文化实践和身份认同，而这通常通过实现下载的应用程序或在线服务来实现。从社交媒体平台上填写的对话框和提示，到晦涩难懂的信用评分计算，我们几乎无条件地遵循着这些规则。为了深入挖掘算法，我们需要理解计算思维的全部范围及其与程序性语言神话的相互作用，而这首先要从我们对算法应有的理解入手。

This figure of the algorithm as a quasi-mystical structure of implemented knowledge is both pervasive and poorly understood. We have never been closer to making the metaphor of fully implemented computational knowledge real than we are today, when an explosion of platforms and systems is reinventing cultural practice and identity, often by implementing a me downloaded as an app or set up as an online service. We are surrounded by nam-shubs that we obey almost unquestioningly, from the dialog boxes and prompts we fill out on social media platforms to the arcane computation of credit scores. To begin excavating the algorithm, we need to understand the full scope of computational thinking and its interactions with the mythos of procedural language, starting with what we think algorithms ought to be.

计算大教堂

The Cathedral of Computation

如今，当技术专家、研究人员和企业家谈论计算文化时，算法的深层迷思常常被层层合理化的修辞和软件设计的程序性隐喻所掩盖。事实上，最流行的隐喻似乎是将代码比作结构：平台、架构、对象、门户、网关。这既降低了软件的人格，淡化了软件主体的概念（建筑物是被动的；行动的是建筑师、工程师和用户），又将代码具体化为一种客观的构造，就像建筑物一样，存在于世界中。

When technologists, researchers, and entrepreneurs speak about computational culture today, this deep myth of the algorithm is typically obscured by layers of rationalizing rhetoric and the procedural metaphors of software design. Indeed the most prevalent set of metaphors seems to be that of code as structure: platforms, architectures, objects, portals, gateways. This serves to both depersonify software, diluting the notion of software agency (buildings are passive; it’s the architects, engineers, and users who act), and reifying code as an objective construct, like a building, that exists in the world.

然而，即使在这种建筑语言中，算法的神话形象仍然重新出现。考虑将大教堂作为代码隐喻的流行程度。乔治戴森关于计算兴起的精彩历史著作名为《图灵大教堂》。另一个经典实例是埃里克雷蒙德关于开源软件开发的书《大教堂与集市》（雷蒙德主张更透明的集市模式，而不是大教堂自上而下的方法）。但也许最好的比喻是 1988 年 IEEE 计算机协会提出的：“软件和大教堂非常相似——我们先建造它们，然后我们祈祷。” ⁷当然，这是一个玩笑，但它隐藏了关于我们与当今算法形象关系的更深层次的真相。代码的体系结构依赖于信念结构以及位的逻辑组织。

Yet even within this architectural language, the mythological figure of the algorithm reasserts itself. Consider the popularity of the cathedral as a metaphor for code. George Dyson’s wonderful history of the rise of computation is titled Turing’s Cathedral. Another classic instantiation is Eric Raymond’s book on open source software development, The Cathedral and the Bazaar (Raymond was arguing for the more transparent bazaar model, rather than the top-down approach of the cathedral). But perhaps the best analogy was offered at the IEEE Computer Society in 1988: “Software and cathedrals are much the same—first we build them, then we pray.”⁷ This was meant as a joke, of course, but it hides a deeper truth about our relationship to the figure of the algorithm today. The architecture of code relies on a structure of belief as well as a logical organization of bits.

大教堂并非计算的完美隐喻，但它的缺陷恰恰表明了我们所忽略的东西。大教堂既是物质的，也是精神的，是上帝的殿堂。从这个意义上说，建筑的物理外观讲述着关于信仰和实践的特定故事（例如，洗礼盆、指向东方的中殿、圣经故事的插图）。但它也暗示了一种通往无形的宗教空间——超越物质现实的上帝殿堂——的特殊方式：变体论、圣物和仪式都是大教堂奇观的一部分，反映了无形的信仰机制。然而，这些机制中的大部分不可避免地隐藏在外：教会分裂、预算、丑闻、教义矛盾，以及其他软件工程师称之为大教堂“后端”的元素，并非其呈现给世人的物质或精神外观的一部分。事实上，当奇观出现片刻停顿，一些令人不安的事实突然浮现在眼前时，正常的本能是忽略它，为了维护自己的信仰而去支撑大教堂的外表。大教堂是一个集体信仰的空间，一个体现对世界理解框架的结构，有些是可见的，有些则不可见。

The cathedral is not a perfect metaphor for computation, but its flaws signal precisely what we are missing. A cathedral is a physical and a spiritual structure, a house of God. In that sense the physical appearance of the building tells particular stories about faith and practice (e.g., a baptismal font, a nave pointing east, illustrations of biblical stories). But it also suggests a particular mode of access to the invisible space of religion, the house of God that exists beyond physical reality: transubstantiation, relics, and ceremonies are all part of the spectacle of the cathedral that reflect the invisible machinery of faith. Yet most of that machinery inevitably remains hidden: schisms, budgets, scandals, doctrinal inconsistencies, and other elements of what a software engineer might call the “back-end” of the cathedral are not part of the physical or spiritual facade presented to the world. Indeed, when the spectacle stutters for a moment and some uncomfortable fact lurches into view, the normal instinct is to ignore it, to shore up the facade of the cathedral in order to maintain one’s faith. A cathedral is a space for collective belief, a structure that embodies a framework of understandings about the world, some visible and some not.

这是一个很贴切的比喻，有助于理解我们如今与算法的关系。 2015年初，数字文化评论家兼游戏设计师伊恩·博格斯特在《大西洋月刊》上发表了一篇题为《计算大教堂》的文章，指出我们与软件之间日益神话化的关系。博格斯特认为，我们已经陷入了一种“计算神权政治”，用算法取代了上帝：

This is a useful metaphor for understanding the relationship we have with algorithms today. Writing in The Atlantic in early 2015, digital culture critic and game designer Ian Bogost called out our increasingly mythological relationship with software in an article titled “The Cathedral of Computation.” Bogost argues that we have fallen into a “computational theocracy” that replaces God with the algorithm:

我们所谓的算法文化与其说是一种物质现象，不如说是一种虔诚的文化，是向人们允许在他们心中取代神的计算机发出的祈求，尽管他们同时声称科学使我们不受宗教的影响^。8

Our supposedly algorithmic culture is not a material phenomenon so much as a devotional one, a supplication made to the computers people have allowed to replace gods in their minds, even as they simultaneously claim that science has made us impervious to religion.⁸

他认为，我们与算法文化机器建立了一种基于信仰的关系，这些机器引导我们穿越城市街道，为我们推荐电影，并为我们提供搜索查询的答案。我们想象这些算法优雅、简洁、高效，但它们却是庞大的集合体，涉及多种形式的人类劳动、物质资源和意识形态选择。

We have, he argues, adopted a faith-based relationship with the algorithmic culture machines that navigate us through city streets, recommend movies to us, and provide us with answers to search queries. We imagine these algorithms as elegant, simple, and efficient, but they are sprawling assemblages involving many forms of human labor, material resources, and ideological choices.

博格斯特的核心论点是：虽然我们将算法视为启蒙运动和理性主义思想的巅峰，但我们与算法的互动却以一种截然不同的方式运作。通过黑匣子、设计简洁的仪表盘以及令人费解的应用程序界面，我们被要求对这种计算深信不疑。正如那些生产高科技产品的低薪工厂工人被隐藏在光滑的拉丝金属制品的设计和营销背后，这些制品似​​乎直接来自某种未经人类之手触碰的机器乌托邦，我们作为这种乌托邦的热切受众，也毫不怀疑地将软件算法的结果视为计算的神奇产物。启蒙运动的商品化是有代价的。它把进步和计算效率变成了一场表演，一场奇观，遮蔽了无所不知的代码神话背后真正的决策和权衡。

Bogost’s central argument is this: while we imagine algorithms as a pinnacle of Enlightenment, rationalist thought, our engagements with them function in a very different mode. Through black boxes, cleanly designed dashboards, and obfuscating Application Program Interfaces, we are asked to take this computation on faith. Just as the poorly paid factory workers who produce our high-tech gadgets are obscured behind the sleek design and marketing of brushed-metal objects that seem to manifest directly from some kind of machine utopia, untouched by human hands, so do we, the eager audience of that utopia, accept the results of software algorithms unquestioningly as the magical products of computation. The commodification of the Enlightenment comes at a price. It turns progress and computational efficiency into a performance, a spectacle that occludes the real decisions and trade-offs behind the mythos of omniscient code.

我们之所以相信这一点，是因为我们长期以来一直生活在算法的神话中——比计算先驱艾伦·图灵甚至查尔斯·巴贝奇及其关于思维机器的推测还要早得多。大教堂在这里是一个普遍的隐喻，因为它提供了一种排序逻辑，一种我们如何组织生活意义的上层建筑或本体论。博格斯特在他的文章中引用了启蒙运动，这是正确的，尽管我认为算法文化与理性主义传统之间的关系远比简单的拒绝或神化更为复杂。我们今天所面临的问题不是我们把计算变成了一座大教堂，而是计算越来越多地取代了一座已经存在的大教堂。这是启蒙运动对普遍知识体系的雄心壮志的大教堂。当我们将两者并列时，我们将我们的信仰投入到一系列已实现的系统中，这些系统承诺为我们完成理性主义的工作，从自动化工厂到自动化科学。

And we believe it because we have lived with this myth of the algorithm for a long time—much longer than computational pioneers Alan Turing or even Charles Babbage and their speculations about thinking machines. The cathedral is a pervasive metaphor here because it offers an ordering logic, a superstructure or ontology for how we organize meaning in our lives. Bogost is right to cite the Enlightenment in his piece, though I will argue the relationship between algorithmic culture and that tradition of rationalism is more complicated than a simple rejection or deification. The problem we are struggling with today is not that we have turned computation into a cathedral, but that computation has increasingly replaced a cathedral that was already here. This is the cathedral of the Enlightenment’s ambitions for a universal system of knowledge. When we juxtapose the two we invest our faith into a series of implemented systems that promise to do the work of rationalism on our behalf, from the automated factory to automated science.

我在第二章中更详细地探讨了这种关系，但现在我们只需理解计算大教堂作为统一理解体系的简写的含义。欧洲大教堂的浅浮雕、雕像和铭文是基督教的缩影，它们概括了福音书和其他重要的圣经叙事，以及它们自身作为持久而完整的信仰陈述的诞生历史。当代计算系统也扮演着同样的角色，通过简洁的界面和精心整理的数据呈现统一的世界观——你想知道的一切，现在都以应用程序的形式呈现。计算提供了一种实现一致性的途径，或者说，将所有知识领域统一到一棵树中：一种信息本体论，其建立在这样一种理念之上：计算是一种通用溶剂，可以解开任何复杂系统，从人类意识到宇宙本身。

I address this relationship more closely in chapter 2, but for now we need only to appreciate the implications of the cathedral of computation as shorthand for a unified system of understanding. The bas-relief work, statues, and inscriptions of great European cathedrals are microcosms of Christianity, recapitulating the Gospel and other key biblical narratives as well as the histories of their own creation as enduring and complete statements of faith. Contemporary computational systems perform the same role of presenting a unified vision of the world through clean interfaces and carefully curated data—everything you might want to know, now available as an app. Computation offers a pathway for consilience, or the unification of all fields of knowledge into a single tree: an ontology of information founded on the idea that computation is a universal solvent that can untangle any complex system, from human consciousness to the universe itself.

数学史学家戴维·柏林斯基 (David Berlinski) 的《算法的出现》是少数几部对算法概念进行长篇研究的作品之一，该书甚至在结论中将通用计算的概念与智能设计联系起来。他认为，算法是“有效计算”概念的透镜，它所做的无非是“使现代世界成为可能”。9^柏林斯基认为“智能在外星海岸的出现”——也就是在计算空间中——进一步证明了在系统本身之外一定存在着对宇宙本质的某种解释。10^他的工作重点是信息和意义之间的区别，图灵机在处理磁带上符号时所做的工作与这些符号对人类思维的影响之间的区别。我们听到了《雪崩》的回响，图灵和他的数学家同事埃米尔·波斯特 (Emil Post) 对通用计算机器的愿景是

One of the few long-form investigations of the algorithm as a concept, mathematical historian David Berlinski’s Advent of the Algorithm, even concludes with an argument connecting the notion of universal computation to intelligent design. He argues that the algorithm, a lens for the notion of “effective calculation,” has done nothing less than to have “made possible the modern world.”⁹ Berlinski sees “the appearance of intelligence on alien shores”—that is, in the spaces of computation—as further evidence that some explanation for the nature of the universe must exist beyond the system itself.¹⁰ His work turns on the distinction between information and meaning, between the work a Turing Machine does in processing symbols on a tape and the impact of those symbols on the human mind. We hear echoes of Snow Crash in the suggestion that Turing and fellow mathematician Emil Post’s visions of universal calculating machines are

对思想的世界做出反应，而不是对物质做出反应。……他们的机器的本质在别处，在一个符号由符号驱动的宇宙中，根据规则，而规则本身就是用符号表达的。

这些机器的居所是人类的思维^。11

responsive to a world of thought, and not matter at all. … The essence of their machines is elsewhere, in a universe in which symbols are driven by symbols according to rules that are themselves expressed in symbols.

The place in which these machines reside is the human mind.¹¹

这正是博格斯特在其文章中呼吁的神化，他指出，我们掩盖了算法的物质现实，将计算视为普遍真理的神秘概念掩盖了起来。我们看到，在算法与文化的交汇处，这种对计算的信仰被反复提及。Facebook 的使命宣言是“赋予人们分享的力量，让世界更加开放和互联”，这一立场包含着一些假设，例如：其社交图谱算法将赋予我们权力；其封闭的专有平台将带来更高的透明度；透明度带来自由，甚至可能带来同理心。Uber 正在“改变世界的出行方式。通过我们的应用程序将乘客与司机无缝连接，我们让城市更加便捷，为乘客开启更多可能性，为司机带来更多业务。”计算的神权政治不仅会改变世界，还会使其发展，并将为用户开启新的可能性，将专有商业与个人自由联系起来。这些变化不仅会影响物质领域，还会影响文化、精神，甚至精神领域的赋权和自主权。该算法为我们提供救赎，但只有在我们接受其服务条款之后。

This is precisely the apotheosis that Bogost calls out in his essay, suggesting that we have veiled the material realities of algorithms behind a mystical notion of computation as a universal truth. We see this faith in computation invoked repeatedly at the intersection of algorithms and culture. Facebook’s mission statement is “to give people the power to share and make the world more open and connected,” a position that embeds assumptions like the argument that its social graph algorithms will grant us power; that its closed, proprietary platform will lead to more transparency; and that transparency leads to freedom, and perhaps to empathy. Uber is “evolving the way the world moves. By seamlessly connecting riders to drivers through our apps, we make cities more accessible, opening up more possibilities for riders and more business for drivers.” The theocracy of computation will not merely change the world but evolve it, and it will open new possibilities for users, linking proprietary commerce and individual freedom. These changes will be effected not only in the material realm but in the cultural, mental, and even spiritual spaces of empowerment and agency. The algorithm offers us salvation, but only after we accept its terms of service.

这里重要的教训不仅仅是硅谷的风险资本主义是资助我们当代大教堂建设的意识形态，甚至不仅仅是博格斯特关于计算是一种新神学的警告。教训在于，当一组思想被组装成一个相互关联的结构时，质疑它们会变得更加困难。一个看似完整一致的知识体系表达，没有任何缝隙，没有任何切入点，表明该结构可能存在外部或替代方案。算法的故事就是差距的故事：理想与实现的计算系统之间，或信息与意义之间的差距。我们为应对生活中日益增长的计算能力而讲述的故事，有时会追溯到这些紧张关系，但它们往往忽略了这种差距，而倾向于一个更舒适或更神奇的世界。正如著名人类学家布罗尼斯瓦夫·马林诺夫斯基近一个世纪前所说：“魔法为原始人提供了许多现成的仪式行为和信仰，以及一种明确的精神和实践技巧，用来弥合每一个重要追求或危急处境中的危险差距。” ¹²面对计算技术的快速进步，我们似乎始终感觉自己更加原始。

The important lesson here is not merely that the venture capitalism of Silicon Valley is the ideology bankrolling much of our contemporary cathedral-building, or even Bogost’s warning about computation as a new theology. The lesson is that it’s much harder to question a set of ideas when they are assembled into an interconnected structure. A seemingly complete and consistent expression of a system of knowledge offers no seams, no points of access that suggest there might be an outside or alternative to the structure. The story of algorithms is the story of the gap: the space between ideal and implemented computational systems, or between information and meaning. The stories we tell ourselves to deal with the growing power of computation in our lives sometimes trace these tensions, but often they ignore the gap in favor of a more comfortable, or a more magical world. As the famed anthropologist Bronisław Malinowski put it almost a century ago, “magic supplies primitive man with a number of ready-made ritual acts and beliefs, with a definite mental and practical technique which serves to bridge over the dangerous gaps in every important pursuit or critical situation.”¹² In the face of computation’s rapid advancement, we seem to feel more primitive all the time.

本书并非对算法的论证，而是一张新的地图，它基于算法作为实现计算理念的工具的重要作用，引领我们探索这一领域。简而言之，算法跨越了鸿沟。它由计算和人类智力材料汇编而成，不断协调计算与物质现实之间的张力。它投射出双重阴影，分别被数理逻辑和文化理解的光芒照亮。本书旨在描绘这一千变万化理念的轮廓。

This book is not an argument for or against the algorithm, but rather a new map to the territory grounded in its vital role as a tool that implements computational ideas. Quite simply, the algorithm spans the gap. Compiled from computational and human intellectual materials, it constantly negotiates the tensions between computation and material reality. It casts a double shadow, illuminated on different sides by the light of mathematical logic and cultural understanding, and this book is an effort to trace the silhouette of this protean idea.

本书的计划

Program of the Book

我们首先从算法的思想根源到其当代文化存在，进行概述。第一章从四个知识线索全面解读算法这一关键概念，首先探讨算法在计算机科学中的基础以及“有效可计算性”的概念。第二条线索探讨控制论以及关于具身化、抽象化和信息论的持续争论。第三条线索，我将回归魔法及其与象征主义的交叠，探讨软件、“源码”以及隐喻表征现实的力量等概念。第四条线索，我将追溯技术性的悠久历史以及人类与文化工具的共同进化。综合这些线索，我将在流程和实施的背景下，将算法定义为文化机器。本章最后总结了算法解读的基本方面，并简要概述了算法的想象力。

We begin with a sketch of the algorithm from its intellectual taproots to its contemporary presence in culture. Chapter 1 lays out a full reading of the algorithm as a critical concept across four intellectual strands, beginning with its foundations in computer science and the notion of “effective computability.” The second strand considers cybernetics and ongoing debates about embodiment, abstraction, and information theory. Third, I return to magic and its overlap with symbolism, engaging with notions of software, “sourcery,” and the power of metaphors to represent reality. Fourth, I draw in the long history of technicity and humanity’s coevolution with our cultural tools. Synthesizing these threads, I offer a definition of the algorithm as culture machine in the context of process and implementation. The chapter closes with a summary of the essential facets of algorithmic reading and a brief glimpse of algorithmic imagination.

接下来我们转到由 DARPA 资助的智能助手（现称为 Siri）项目的介绍。第二章探讨了谷歌、苹果和其他公司如何将文化算法的发展转化为对自我认知和普遍知识的认识论探索。Siri 承诺的预期性、亲密性知识，在斯派克·琼斯的电影《她》中发挥到了极致，它试图比任何情人都更彻底地描绘出个人心灵的内心空间。这种诱惑与谷歌打造“星际迷航计算机”的努力形成了鲜明的对比，后者可以使用其完美的共享外部知识地图来回答任何问题。这些探索既浪漫又理性，寻求一种超越认知的境界，这种境界只有通过最终超越人类的机制才能达到。科技巨头们雄心勃勃，致力于开发能够“回答、对话和预测”的算法，他们正在塑造我们算法的未来，并构建一个关于可知和可欲事物的全新认识论框架：一个知识需求层次，它最终不仅将勾勒出公共信息领域，还将勾勒出人类身份认同的内在空间。我将这种对知识的追求与其伟大的启蒙先驱——哲学家兼出版商丹尼斯·狄德罗创作的第一部百科全书——并列，追溯两部为完善知识而构建的文化机器之间的相似之处。

From there we move to an account of the DARPA-funded project to create an intelligent assistant (now known as Siri). Chapter 2 explores the ways in which Google, Apple, and other corporations have turned the development of cultural algorithms into epistemological quests for both self-knowledge and universal knowledge. The anticipatory, intimate knowledge promised by Siri, taken to its logical extreme in the Spike Jonze film Her, is an attempt to map out the inner space of the individual psyche more thoroughly than any lover could. This seduction contrasts neatly with Google’s drive to create a “Star Trek computer” that can answer any question using its perfect map of shared, external knowledge. These quests are both romantic and rational, seeking a transcendent state of knowing, a state that can be reached only with mechanisms that ultimately eclipse the human. Through their ambitions to develop algorithms that can “answer, converse and anticipate,” the technology titans shaping our algorithmic future are constructing a new epistemological framework of what is knowable and desirable: an intellectual hierarchy of needs that will ultimately map out not only the public sphere of information but the interior space of human identity. I juxtapose this quest for knowledge with its great enlightenment precursor, the creation of the first Ecyclopédie by philosopher-publisher Denis Diderot, tracing the parallels in the construction of two culture machines for the perfection of knowledge.

在确定了算法知识的出现之后，故事在第三章转向了算法美学的兴起。Netflix 是我们的一个陪衬：它拒绝使用大数据统计方法来评判品味，转而采用混合人机计算模型，这导致了博尔赫斯式的项目，将所有真实和潜在的电影分类为 76,897 个类型之一。这个庞大的分析企业塑造了该公司对原创作品的创意投资，尤其是其原创电视剧《纸牌屋》。我通过该剧的开发和发行故事来论证文化的算法模型正日益具有影响力且不可避免。该剧的情感和商业地位让我们得以一窥新的算法美学，这些美学以相互竞争的方式个性化和单一化。我在本章的最后指出，Netflix 通过操纵某些类型的计算抽象来实现文化和财务上的成功，展示了文化套利的力量和陷阱。

Having established the emergence of algorithmic knowledge, the story turns to the rise of algorithmic aesthetics in chapter 3. Netflix serves as our foil: its rejection of a big-data statistics approach to taste in favor of a hybrid human-computational model has led to the Borges-esque project of taxonomizing all real and potential films into one of 76,897 genres. This massive analytical enterprise shapes the company’s creative investments in original work, particularly its original television series House of Cards. I use the story of the show’s development and distribution to argue that algorithmic models of culture are increasingly influential and inescapable. The affective and commercial position of the series offers us a glimpse of new algorithmic aesthetics that are personalized and monolithic in competing ways. I close the chapter by arguing that Netflix demonstrates the power and pitfalls of cultural arbitrage by manipulating certain kinds of computational abstraction to achieve cultural and financial success.

随着算法越来越擅长解读文化数据并进行实时套利（此处指金融定价套利，也指上一章所述的文化套利），它们正在接管新的智力劳动形式。它们在创作的同时，也进行着简化和抽象，在消费者与诸如打车或雇佣保姆之类的繁琐流程之间构建了一个接口层。第四章以伊恩·博格斯特（Ian Bogost）的讽刺性Facebook游戏《奶牛点击者》（Cow Clicker ）及其对“游戏化”运动的戏仿开篇，该运动旨在将量化和算法思维融入日常生活的方方面面。这类游戏挑战了工作与娱乐之间的界限，就像Uber和高科技仓库工人那样，他们的每一秒、每一步都被衡量以衡量效率，这种更为严肃的游戏化形式也同样如此。总而言之，这些新的工作模式预示着算法时代一种新型的异化劳动。在我们的科幻小说中，人类是处理器，负责处理算法装置分配的简单任务。通过借鉴自动机这一历史人物、由 Mechanical Turk 驱动的诗歌选集《分包合同》以及亚当·斯密在《道德情操论》中对同理心的概念，我探讨了计算资本主义对政治、同理心和社会价值的影响。

As algorithms become more adept at reading cultural data and performing real-time arbitrage (used here in the sense of financial pricing arbitrage but also cultural arbitrage as described in the previous chapter), they are taking on new forms of intellectual labor. They are authoring and creating, but they are also simplifying and abstracting, creating an interface layer between consumers and the messy process of, say, getting a cab or hiring a housekeeper. Chapter 4 begins with Ian Bogost’s satirical Facebook game Cow Clicker and its send-up of the “gamification” movement to add quantification and algorithmic thinking to many facets of everyday life. Such games trouble the boundaries between work and play, as do much more serious forms of gamification like Uber and the high-tech warehouse workers whose every second and step are measured for efficiency. Taken together, these new models of work herald a novel form of alienated labor for the algorithmic age. In our science fiction present, humans are processors handling simple tasks assigned by an algorithmic apparatus. Drawing on the historical figure of the automaton, a remarkable collection of Mechanical Turk-powered poetry titled Of the Subcontract, and Adam Smith’s conception of empathy in his Theory of Moral Sentiments, I explore the consequences of computational capitalism on politics, empathy, and social value.

算法巨变的根源在于以计算的方式重新构想价值。第五章以2010年的闪电崩盘以及算法交易在国际市场日益增长的主导地位（记者迈克尔·刘易斯的《闪速小子》等作品对此进行了描述）为开端，以此构建对比特币及相关加密货币的解读。通过计算周期定义交易单位，比特币从根本上将基于信仰的货币共同体从物质主义价值体系转变为算法价值体系。从脸书到新闻业，算法套利正在推动许多文化交流领域在价值和意义的归属上发生类似的转变。从重视文化对象本身到重视该对象所建立或支持的关系网络的根本转变，带来了新的生产实践和美学，形式和类型让位于模因和模糊的合作作品。以比特币作为这种新价值模型的例子，我最后探讨了可编程价值对21世纪公共领域概念的影响，在这个时代，套利胜过内容。

The root of the algorithmic sea change is the reimagination of value in computational terms. Chapter 5 leads with the flash crash in 2010 and the growing dominance of algorithmic trading in international markets (described by journalist Michael Lewis’s Flash Boys, among others) to frame a reading of Bitcoin and related cryptocurrencies. By defining the unit of exchange through computational cycles, Bitcoin fundamentally shifts the faith-based community of currency from a materialist to an algorithmic value system. Algorithmic arbitrage is forcing similar transitions in the attribution of value and meaning in many spaces of cultural exchange, from Facebook to journalism. The fundamental shift from valuing the cultural object itself to valuing the networks of relations that the object establishes or supports leads to new practices and aesthetics of production, where form and genre give way to memes and nebulous collaborative works. Using Bitcoin as an example of this new value model, I close by considering the consequences of programmable value for the notion of a public sphere in the twenty-first century, an era when arbitrage trumps content.

在尾声中，我简要回顾了算法的谱系，以思考我们实现有效可计算性核心中蕴含的双重愿望的未来前景：对普遍知识的追求和对完美自我的认知。这些抱负对人文学科尤为重要，我们不能止步于算法解读。为了真正应对算法时代以及我们与计算文化进程日益深化的纠缠，我们需要作为学者、教师，尤其是人文探究的实践者采取行动。我们需要一种实验性人文学科，一套直接参与算法生产和学术研究的策略，借鉴即兴创作和实验研究的理论，论证一种过程文化，即算法生产的文化，需要一种既反思又充满趣味的过程性批判。如此，我们才能开始将算法的形象理解为文化想象空间的重新绘制，并成为文化机器的真正合作者，而不是它们的崇拜者，甚至更糟的是，成为它们的宠物。

In the coda I briefly retrace this genealogy of the algorithm to consider our future prospects for achieving the twinned desires embedded in the heart of effective computability: the quest for universal knowledge and perfect self-knowledge. These ambitions are particularly vital for the humanities, and we cannot stop at algorithmic reading. To truly grapple with the age of the algorithm and our growing entanglement with computational cultural processes, we need to take action as scholars, teachers, and most of all performers of humanistic inquiry. We need an experimental humanities, a set of strategies for direct engagement with algorithmic production and scholarship, drawing on theories of improvisation and experimental investigation to argue that a culture of process, of algorithmic production, requires a processual criticism that is both reflexive and playful. This is how we can begin to understand the figure of the algorithm as a redrawing of the space for cultural imagination and become true collaborators with culture machines rather than their worshippers or, worse, their pets.

笔记

Notes

文化机器的崛起

Rise of the Culture Machines

21世纪初，我们与计算机的关系发生了变化。我们开始把设备放在口袋里，在餐桌上仔细端详它们，在角落里轻声低语。我们不再思考硬件，而是开始思考应用程序和服务。我们不仅使用计算系统，而且信任它们，它们会告诉我们去哪里、和谁约会、思考什么（仅举几个例子）。每一次点击，每一份服务协议条款，都让我们相信大数据、无处不在的传感器以及各种形式的机器学习可以建模并有效地管理各种复杂系统，从挑选歌曲到预测犯罪。在此过程中，一个古老的词再次焕发新生：算法。无论是被忽视还是被过度炒作，“算法”很少被认真地视为计算机为我们所做的文化工作中的一个关键术语。本书将“算法”一词拆解开来，又重新组合起来，展示算法如何作为我们需要学习如何阅读和理解的文化机器发挥作用。

Sometime in the late 2000s, our relationship with computers changed. We began carrying devices around in our pockets, peering at them at the dinner table, muttering quietly to them in the corner. We stopped thinking about hardware and started thinking about apps and services. We have come not just to use but to trust computational systems that tell us where to go, whom to date, and what to think about (to name just a few examples). With every click, every terms of service agreement, we buy into the idea that big data, ubiquitous sensors, and various forms of machine learning can model and beneficially regulate all kinds of complex systems, from picking songs to predicting crime. Along the way, an old word has become new again: the algorithm. Either overlooked or overhyped, the algorithm is rarely taken seriously as a key term in the cultural work that computers do for us. This book takes that word apart and puts it back together again, showing how algorithms function as culture machines that we need to learn how to read and understand.

算法无处不在。它们已经主宰了股票市场、作曲、驾驶汽车、撰写新闻文章，并撰写冗长的数学证明——而它们创造性的写作能力才刚刚开始显现。各大公司小心翼翼地守护着运行这些数据和流程集合的黑匣子。即使是世界上一些最成功、最普遍的算法系统背后的工程师——例如谷歌和Netflix的高管——也承认，他们只理解系统所展现的部分行为。但他们的言论仍然具有超越性和解放性，当他们将计算与变革性的正义和自由等同起来时，其技术乌托邦的基调与代码即魔法的神话如出一辙。伊恩·博格斯特所定义的计算神学是一位信仰斗士，将大数据和颠覆性的福音带给了社会的广大领域。

Algorithms are everywhere. They already dominate the stock market, compose music, drive cars, write news articles, and author long mathematical proofs—and their powers of creative authorship are just beginning to take shape. Corporations jealously guard the black boxes running these assemblages of data and process. Even the engineers behind some of the most successful and ubiquitous algorithmic systems in the world—executives at Google and Netflix, for example—admit that they understand only some of the behaviors their systems exhibit. But their rhetoric is still transcendent and emancipatory, striking many of the same techno-utopian notes as the mythos of code as magic when they equate computation with transformational justice and freedom. The theology of computation that Ian Bogost identified is a faith militant, bringing the gospel of big data and disruption to huge swaths of society.

这就是我们如今使用算法的语境：算法如同日常的技术魔法，我们将其用于预订假期、推荐潜在伴侣、评估标准化考试论文以及执行许多其他类型的文化工作。华尔街交易员们将他们的金融“算法”命名为“伏击”和“袭击者”，但他们往往并不清楚这些赚钱的黑匣子是如何运作的。2^作为文化评论家雷蒙德·威廉姆斯（Raymond Williams）精神中的关键词，³ “算法”一词通常涵盖一系列计算过程，包括对用户行为的密切监控、对结果信息的“大数据”聚合、结合多种统计计算形式来解析数据的分析引擎，以及最终一系列面向人类的操作、建议和界面，这些通常只反映了幕后文化处理的一小部分。计算在世界中逐渐拥有某种存在，成为一种“事物”，它既掩盖了温迪·惠京·全（Wendy Hui Kyong Chun）所说的“可编程性”的特定形式，也突显了这种形式——我们将在下文中以计算主义的视角重新讨论这一概念^。4

This is the context in which we use algorithms today: as pieces of quotidian technical magic that we entrust with booking vacations, suggesting potential mates, evaluating standardized test essays, and performing many other kinds of cultural work. Wall Street traders give their financial “algos” names like Ambush and Raider, yet they often have no idea how their money-making black boxes work.² As a keyword in the spirit of cultural critic Raymond Williams,³ the word algorithm frequently encompasses a range of computational processes including close surveillance of user behaviors, “big data” aggregation of the resulting information, analytics engines that combine multiple forms of statistical calculation to parse that data, and finally a set of human-facing actions, recommendations, and interfaces that generally reflect only a small part of the cultural processing going on behind the scenes. Computation comes to have a kind of presence in the world, becoming a “thing” that both obscures and highlights particular forms of what Wendy Hui Kyong Chun calls “programmability,” a notion we will return to in the guise of computationalism below.⁴

正是计算的这种千变万化的本质既困扰着我们，又吸引着我们。有时，计算系统似乎符合那种离散的“物性”标准，就像苏美尔神话中的“我”或智能手机屏幕上闪亮的应用程序按钮。有时，它们更难与更广泛的文化环境区分开来：拼写检查程序在多大程度上通过数十亿次细微的修正改变了措辞和语法选择？我们如何理清构成它们底层的代码、词典和语法的集合？虽然计算的文化效应及其影响非常复杂，但这些系统通过人类设计和实施的工具在世界上发挥作用。为了建立一个解读文化计算的关键框架，我们必须从这些挤在算法这个简陋容器中的工具开始。

It is precisely this protean nature of computation that both troubles and attracts us. At some times computational systems appear to conform to that standard of discrete “thingness,” like the me of Sumerian myth or a shiny application button on a smartphone screen. At other moments they are much harder to distinguish from broader cultural environments: to what extent are spell-check programs changing diction and grammatical choices through their billions of subtle corrections, and how do we disentangle the assemblage of code, dictionaries, and grammars that underlie them? While the cultural effects and affects of computation are complex, these systems function in the world through instruments designed and implemented by human beings. In order to establish a critical frame for reading cultural computation, we have to begin with those instruments, jammed together in the humble vessel of the algorithm.

我们对《雪崩》的观察揭示了层层魔法、“巫术”和结构化信仰，它们构成了当今文化中算法表象的基石。现在，我们来看看那些实现计算系统的工程师和计算机科学家。这种算法植根于计算机科学，并依赖于数学史。算法是一种处方、一套指令、一系列用于实现特定计算或结果的任务，例如计算平方根或编制斐波那契数列所需的步骤。“算法”一词源于公元九世纪著名的数学家阿布·阿卜杜拉·穆罕默德·伊本·穆萨·花拉子密（代数也源于他的名字）。“算法”（Algorismus ）最初是指计算印度-阿拉伯数字的过程。通过花拉子密，该算法与位置表示法、小数点和零等革命性概念联系在一起。

Our look at Snow Crash revealed the layers of magic, “sourcery,” and structured belief that underpin the facade of the algorithm in culture today. Now we turn to the engineers and computer scientists who implement computational systems. Rooted in computer science, this version of the algorithm relies on the history of mathematics. An algorithm is a recipe, an instruction set, a sequence of tasks to achieve a particular calculation or result, like the steps needed to calculate a square root or tabulate the Fibonacci sequence. The word itself derives from Abū ʿAbdallāh Muḥammad ibn Mūsā al-Khwārizmī, the famed ninth-century CE mathematician (from whose name algebra is also derived). Algorismus was originally the process for calculating Hindu-Arabic numerals. Via al-Kwarizmi, the algorithm was associated with the revolutionary concepts of positional notation, the decimal point, and zero.

随着“算法”一词在随后的几个世纪中逐渐流行，“算法”一词开始用来描述任何一组用于处理数据或推理问题的数学指令。巴比伦人使用了一些最早的数学算法来推导平方根和因数。^欧几里得设计了一种算法，用于求两个数的最大公约数。在整个演化过程中，该算法保留了一个即将成为核心特性的基本特性：它就是有效。也就是说，算法能够在有限的时间内可靠地提供预期结果（或许，除了那些让数学家着迷、让工程师恼火的极端情况）。

As the word gained currency in the centuries that followed, “algorithm” came to describe any set of mathematical instructions for manipulating data or reasoning through a problem. The Babylonians used some of the first mathematical algorithms to derive square roots and factor numbers.⁵ Euclid devised an algorithm for taking two numbers and finding the greatest common divisor they share. Throughout this evolution, the algorithm retained an essential feature that will soon become central to the story: it just works. That is to say, an algorithm reliably delivers an expected result within a finite amount of time (except, perhaps, for those edge cases that fascinate mathematicians and annoy engineers).

历史学家内森·恩斯门格 (Nathan Ensmenger) 讲述了计算机科学学科是如何在其倡导者接受算法概念之后才得以凝聚的。该领域的创始人之一唐纳德·克努斯 (Donald Knuth) 在其开创性教科书《计算机编程艺术》中将该领域的起源追溯到花拉子米 (al-Khwarizmi) ^。6算法是一个理想的研究对象，既易于掌握，又充满无穷的谜题：

Historian Nathan Ensmenger recounts how the academic discipline of computer science coalesced only after its advocates embraced the concept of the algorithm, with one of the field’s founders, Donald Knuth, tracing the field’s origins to al-Khwarizmi in his seminal textbook The Art of Computer Programming.⁶ The algorithm was an ideal object of study, both easily grasped and endlessly puzzling:

通过提出算法对于计算技术活动的基础作用，就如同艾萨克·牛顿的运动定律对于物理学一样，克努斯和他的计算机科学家同事们可以宣称自己与广大科学家群体建立了完全的友谊^。7

By suggesting that the algorithm was as fundamental to the technical activity of computing as Sir Isaac Newton’s laws of motion were to physics, Knuth and his fellow computer scientists could claim full fellowship with the larger community of scientists.⁷

然而，正如数学家 Yiannis Moschovakis 指出的那样，Knuth 关于算法究竟是什么的论点是一个极其罕见的例子，它突出了这个问题。8^对于计算机科学家来说，这个术语仍然更多的是一个直观的、未经检验的概念，而不是一个基于数学计算理论的、划定的逻辑概念。

And yet, as mathematician Yiannis Moschovakis points out, Knuth’s argument about what algorithms actually are is an extremely rare instance where the question is foregrounded.⁸ For computer scientists the term remains more of an intuitive, unexamined notion than a delineated logical concept grounded in a mathematical theory of computation.

在很大程度上，多亏了 Knuth，算法如今已成为计算机科学的一个基本概念，是本科生入门“算法与数据结构”课程中重要的知识基石。算法代表着可重复、实用的解决方案，例如将一个数分解成其最小的素数分量，或者找到网络中最高效的路径。当代算法研究的重点并非算法是否有效，而是算法的效率，以及在 CPU 周期、内存和准确性方面需要做出哪些权衡。

Thanks in large part to Knuth, the algorithm today is a fundamental concept in computer science, an intellectual keystone typically covered in the introductory Algorithms and Data Structures course for undergraduate majors. Algorithms represent repeatable, practical solutions to problems like factoring a number into its smallest prime number components or finding the most efficient pathway through a network. The major focus for contemporary algorithmic research is not whether they work but how efficiently, and with what tradeoffs in terms of CPU cycles, memory, and accuracy.

我们可以将这种实用主义的算法方法提炼为一张 PowerPoint 幻灯片。计算算法领域的顶尖研究员罗伯特·塞奇威克 (Robert Sedgewick) 碰巧也教授了我本科时选修的《算法与数据结构》课程；在他广为流传的课程材料中，他把算法称为“解决问题的方法”。9^这就是我所说的实用主义者的定义：工程师对算法的看法是，算法旨在定义问题和解决方案。实用主义者的定义以实用性为基础来宣称其真实性：算法适用于某一目的，照亮问题和解决方案之间的路径。这是主导谷歌、苹果、亚马逊和其他行业巨头工程师的分组讨论室和工作站的关键框架。正如谷歌所描述的：“算法是将您的问题转化为答案的计算机过程和公式。” ¹⁰对于许多工程师和技术人员来说，算法仅仅是工作，是他们劳动的媒介。

We can distill this pragmatic approach to algorithms down to a single PowerPoint slide. Robert Sedgewick, a leading researcher on computational algorithms, also happened to teach the version of Algorithms and Data Structures that I took as an undergraduate; he calls the algorithm a “method for solving a problem” in his widely circulated course materials.⁹ This is what I term the pragmatist’s definition: an engineer’s notion of algorithms geared toward defining problems and solutions. The pragmatist’s definition grounds its truth claim in utility: algorithms are fit for a purpose, illuminating pathways between problems and solutions. This is the critical frame that dominates the breakout rooms and workstations of engineers at Google, Apple, Amazon, and other industry giants. As Google describes them: “Algorithms are the computer processes and formulas that take your questions and turn them into answers.”¹⁰ For many engineers and technologists, algorithms are quite simply the work, the medium of their labor.

实用主义的定义揭示了算法的本质政治性，以及它与工具理性意识形态的明显共谋，数字文化学者大卫·哥伦比亚（David Golumbia）在其计算批判中指出了这一点。11^当然，这正是算法的作用：它们是方法，继承了从阿基米德到万尼瓦尔·布什的科学方法和工程学的归纳传统。它们解决的是那些开发和优化代码的工程师和企业家们所发现的问题。但这样的实现绝不仅仅是代码：解决问题的方法不可避免地涉及各种技术和智力推理、干预和过滤。

The pragmatic definition lays bare the essential politics of the algorithm, its transparent complicity in the ideology of instrumental reason that digital culture scholar David Golumbia calls out in his critique of computation.¹¹ Of course this is what algorithms do: they are methods, inheriting the inductive tradition of the scientific method and engineering from Archimedes to Vannevar Bush. They solve problems that have been identified as such by the engineers and entrepreneurs who develop and optimize the code. But such implementations are never just code: a method for solving a problem inevitably involves all sorts of technical and intellectual inferences, interventions, and filters.

举个例子，思考一下经典的计算机科学问题——旅行商问题：如何计算出一条穿过目的地之间距离各异的地理区域的高效路线？这个问题在现实世界中有很多类似的例子，例如为UPS司机规划路线。事实上，该公司已经投资了数亿美元，开发了一个名为ORION的长达1000页的算法，其决策部分基于旅行商启发式算法。12^然而，旅行商问题将每个目的地想象成图上的一个相同点，而UPS的送货点在完成送货所需的时间上差异很大（例如，用手推车拖着沉重的包裹上路，或者避开店主的猎犬）。ORION的宇宙算法模型必须在特定的计算抽象（每个站点都是一个无特征的可替代点）、人类驾驶员的实际经验和反馈，以及公司收集的有关全球停车标志、转弯车道等状况的数据之间取得平衡。通过网络优化路径的计算机科学问题必须与驾驶员的自主性、对微观逻辑决策（例如是否右转或左转）施加量化跟踪以及其他复杂人类系统（从交通堵塞到宠物）的意外干预共享计算阶段。

As an example, consider the classic computer science problem of the traveling salesman: how can one calculate an efficient route through a geography of destinations at various distances from one another? The question has many real-world analogs, such as routing UPS drivers, and indeed that company has invested hundreds of millions of dollars in a 1,000-page algorithm called ORION that bases its decisions in part on traveling salesman heuristics.¹² And yet the traveling salesman problem imagines each destination as an identical point on a graph, while UPS drop-offs vary greatly in the amount of time they take to complete (hauling a heavy package up with a handcart, say, or avoiding the owner’s terrier). ORION’s algorithmic model of the universe must balance between particular computational abstractions (each stop is a featureless, fungible point), the lived experience and feedback of human drivers, and the data the company has gathered about the state of the world’s stop signs, turn lanes, and so on. The computer science question of optimizing paths through a network must share the computational stage with the autonomy of drivers, the imposition of quantified tracking on micro-logistical decisions like whether to make a right or left turn, and the unexpected interventions of other complex human systems, from traffic jams to pets.

当然，ORION 及其长达 1000 页的针对这一复杂问题的“解决方案”是一个持续演进的过程或系统，而非棕色卡车芭蕾舞般协调的优雅方程式。其方程式和人类行为计算模型只是数百万个试图规范化和优化复杂文化系统的算法中的一个例子。实用主义者的定义通过构建一座隐性知识的大厦（大教堂）来实现清晰度，其中大部分知识都分层存在于抽象系统中，例如旅行商问题。在文化取得成功达到一定程度时，这些系统也开始创造自己的现实：系统中的各种参与者开始以某种方式改变其行为，从而绕过系统的假设。互联网讨论区收集了关于送货司机不敲门而是留下门牌声称住户不在家的投诉。这些快捷方式之所以有效，正是因为它们对于 ORION 等系统来说是不可见的，这使得驾驶员可以节省宝贵的时间，并且可能在日程安排开始延误的忙碌日子里赶上正在跟踪的所有其他指标。

ORION and its 1,000-page “solution” to this tangled problem is, of course, a process or system in continued evolution rather than an elegant equation for the balletic coordination of brown trucks. Its equations and computational models of human behavior are just one example among millions of algorithms attempting to regularize and optimize complex cultural systems. The pragmatist’s definition achieves clarity by constructing an edifice (a cathedral) of tacit knowledge, much of it layered in systems of abstraction like the traveling salesman problem. At a certain level of cultural success, these systems start to create their own realities as well: various players in the system begin to alter their behavior in ways that short-circuit the system’s assumptions. Internet discussion boards catalog complaints about delivery drivers who do not bother to knock and instead leave door tags claiming that the resident was not at home. These shortcuts work precisely because they are invisible to systems like ORION, allowing the driver to save valuable seconds and perhaps catch up on all those other metrics that are being tracked on a hectic day when the schedule starts to slip.

当今许多最强大的公司本质上都是复杂算法的文化包装，我们将在接下来的章节中看到这一点。谷歌就体现了一家建立在PageRank算法之上的公司，甚至是一种完整的世界观。亚马逊的转型算法不仅涉及计算，还涉及物流，它找到了外包、超越传统书商（以及后来几乎所有消费品的销售商）并超越其销量的方法。Facebook开发了世界上最成功的社交算法，用于连接人们。这些只是一些强大、务实、利润丰厚的算法的例子，这些算法不断更新和修改，以应对它们试图计算的复杂文化空间。

Many of the most powerful corporations in existence today are essentially cultural wrappers for sophisticated algorithms, as we will see in the following chapters. Google exemplifies a company, indeed an entire worldview, built on an algorithm, PageRank. Amazon’s transformational algorithm involved not just computation but logistics, finding ways to outsource, outmaneuver, and outsell traditional booksellers (and later, sellers of almost every kind of consumer product). Facebook developed the world’s most successful social algorithm for putting people in contact with one another. These are just a few examples of powerful, pragmatic, lucrative algorithms that are constantly updated and modified to cope with the messy cultural spaces they attempt to compute.

我们生活在一个由算法实用主义者构建的世界里。事实上，像谷歌这样的企业，其雄心壮志和运营规模意味着，他们对算法的定义——问题是什么，以及如何解决——能够深刻地改变世界。他们各种各样实用主义的表达方式，激发了我们精心设计的应对措施和应对方案，或者被传播学研究员塔尔顿·吉莱斯皮称为我们为了适应算法系统而进行的“默契协商”：我们在与机器对话时使用不同的发音，使用主题标签使更新内容更易于机器阅读，并使用搜索引擎友好的术语描述我们的工作^。13

We live, for the most part, in a world built by algorithmic pragmatists. Indeed, the ambition and scale of corporate operations like Google means that their definitions of algorithms—what the problems are, and how to solve them—can profoundly change the world. Their variations of pragmatism then inspire elaborate responses and counter-solutions, or what communication researcher Tarleton Gillespie calls the “tacit negotiation” we perform to adapt ourselves to algorithmic systems: we enunciate differently when speaking to machines, use hashtags to make updates more machine-readable, and describe our work in search engine-friendly terms.¹³

实用主义者定义之下潜藏的默认假设越来越难以忽视。计算系统表面上的透明性和简单性，使许多人将其视为公正决策的工具。UpStart 和 ZestFinance 等公司将计算视为判断金融可靠性的一种方式，并向未通过信用评分等更传统的信用算法测试的人提供贷款。14^这些系统本质上部署算法是为了抵消其他算法的偏见，或者更愤世嫉俗地说，是为了识别其他算法错过的商机。然而，这些系统背后的公司相对不寻常，他们承认其商业计划的意识形态框架，并明确说明其系统如何尝试判断“性格”。

The tacit assumptions lurking beneath the pragmatist’s definition are becoming harder and harder to ignore. The apparent transparency and simplicity of computational systems are leading many to see them as vehicles for unbiased decision-making. Companies like UpStart and ZestFinance view computation as a way to judge financial reliability and make loans to people who fail more traditional algorithmic tests of credit-worthiness, like credit scores.¹⁴ These systems essentially deploy algorithms to counter the bias of other algorithms, or more cynically to identify business opportunities missed by others. The companies behind these systems are relatively unusual, however, in acknowledging the ideological framing of their business plans, and explicitly addressing how their systems attempt to judge “character.”

但如果这些反身性算法旨在利用系统性不平等，那么它们实际上是在回应通常缺乏这种意识的更广泛的文化体系。计算转向意味着，许多算法现在根据隐藏在公众视野之外的数学规则和隐含假设，重建并抹去法律、伦理和感知现实。正如法律伦理学家弗兰克·帕斯夸莱（Frank Pasquale）在谈到评估求职者的算法时所写：

But if these are reflexive counter-algorithms designed to capitalize on systemic inequities, they are responding to broader cultural systems that typically lack such awareness. The computational turn means that many algorithms now reconstruct and efface legal, ethical, and perceived reality according to mathematical rules and implicit assumptions that are shielded from public view. As legal ethicist Frank Pasquale writes about algorithms for evaluating job candidates:

自动化系统声称会以相同的方式对所有个体进行评分，从而避免歧视。它们或许能确保一些老板不再基于直觉、印象或偏见来做出招聘和解雇决定。但软件工程师构建了评分系统挖掘的数据集；他们定义了数据挖掘分析的参数；他们创建了应用的集群、链接和决策树；他们生成了应用的预测模型。人类的偏见和价值观深深地嵌入到开发的每个步骤中。计算机化可能只会加剧歧视^。15

Automated systems claim to rate all individuals the same way, thus averting discrimination. They may ensure some bosses no longer base hiring and firing decisions on hunches, impressions, or prejudices. But software engineers construct the datasets mined by scoring systems; they define the parameters of data-mining analyses; they create the clusters, links, and decision trees applied; they generate the predictive models applied. Human biases and values are embedded into each and every step of development. Computerization may simply drive discrimination upstream.¹⁵

随着算法深入文化空间，其实用主义的定义正受到更严格的审视，正如帕斯夸莱、哥伦比亚以及越来越多的算法伦理学者所言，这些批判框架拒绝接受问题与解决方案的工程准则。世界上实用主义算法得以蓬勃发展的抽象概念和嵌入式系统殿堂，其根基可以追溯到符号逻辑、计算理论和控制论，在这些理性理念的集合中，我们发现了一个奇特的东西：欲望。

As algorithms move deeper into cultural space, the pragmatic definition gets scrutinized more closely according to critical frames that reject the engineering rubric of problem and solution, as Pasquale, Golumbia, and a growing number of algorithmic ethics scholars have argued. The cathedral of abstractions and embedded systems that allow the pragmatic algorithms of the world to flourish can be followed down to its foundations in symbolic logic, computational theory, and cybernetics, where we find a curious thing among that collection of rational ideas: desire.

从计算到欲望

From Computation to Desire

工程师的问题和解决方案背后的真理主张是什么？或者说，支撑源术技术魔力的哲学是什么？它们依赖于受保护的计算空间，即逻辑的、程序性的、非物质的空间，在这个空间里，记忆和过程按照与物质文化截然不同的规则运作。实用主义者的方法指向并且常常依赖于一种关于宇宙本质的更深层次的哲学主张。我们需要将这种主张理解为“有效可计算性”概念的基础，“有效可计算性”是计算机科学中的一个变革性概念，它推动了当今算法的传播。媒体理论家 N. Katherine Hayles在她的书《我的母亲是一台计算机》中将这种哲学主张称为计算体制^。16这是我有时所说的算法时代的另一个术语：算法作为理解宇宙的本体论结构的时代。我们也可以将其视为“计算主义定义”，它扩展了实用主义者的算法概念，并为谷歌和亚马逊等公司的核心商业模式提供了信息。

What are the truth claims underlying the engineer’s problems and solutions, or the philosophy undergirding the technological magic of sourcery? They depend on the protected space of computation, the logical, procedural, immaterial space where memory and process work according to very different rules from material culture. The pragmatist’s approach gestures toward, and often depends on, a deeper philosophical claim about the nature of the universe. We need to understand that claim as the grounding for the notion of “effective computability,” a transformational concept in computer science that fuels algorithmic evangelism today. In her book My Mother Was a Computer, media theorist N. Katherine Hayles labels this philosophical claim the Regime of Computation.¹⁶ This is another term for what I sometimes refer to as the age of the algorithm: the era dominated by the figure of the algorithm as an ontological structure for understanding the universe. We can also think of this as the “computationalist definition,” which extends the pragmatist’s notion of the algorithm and informs the core business models of companies like Google and Amazon.

计算主义在其较为温和的版本中认为，算法在本体论上并不主张真正描述世界，但在解决特定技术问题方面却非常有效。工程师们对宇宙系统持不可知论的态度；他们只关心如何准确地建模宇宙的某些部分，例如与某些查询最匹配的搜索结果，或者华盛顿州斯波坎市的用户今天可能订购的书籍。正如帕斯夸莱以及从杰伦·拉尼尔到叶夫根尼·莫罗佐夫等众多数字文化批评家所言，即使是工程师对算法的定义中隐含的效率和“足够好”的理性主义主张，也会对政策、文化和日常生活实践产生巨大影响，因为算法近似的妥协和类比往往会抹去他们无法理解的一切^。17

In its softer version, computationalism argues that algorithms have no ontological claim to truly describing the world but are highly effective at solving particular technical problems. The engineers are agnostic about the universe as a system; all they care about is accurately modeling certain parts of it, like the search results that best correspond to certain queries or the books that users in Spokane, Washington, are likely to order today. As Pasquale and a host of other digital culture critics from Jaron Lanier to Evgeny Morozov have argued, even the implicit claims to efficiency and “good-enough” rationalism at the heart of the engineer’s definition of algorithms have a tremendous impact on policy, culture, and the practice of everyday life, because the compromises and analogies of algorithmic approximations tend to efface everything that they do not comprehend.¹⁷

计算修辞的扩展很容易渗透到Hayles所谓的计算主义的“硬性主张”中。在这一论证中，算法不仅仅是以或多或少的准确性描述文化过程：这些过程本身就是计算机器，可以在数学上复制（只要有足够的资金）。根据这一逻辑，计算机能够以任何所需的精度模拟选举结果或未来股票价格只是时间和应用科学的问题。计算机科学家和博学家Stephen Wolfram在他雄心勃勃的二十年著作《一种新科学》中阐述了这一论点：

The expansion of the rhetoric of computation easily bleeds into what Hayles calls the “hard claim” for computationalism. In this argument algorithms do not merely describe cultural processes with more or less accuracy: those processes are themselves computational machines that can be mathematically duplicated (given enough funding). According to this logic it is merely a matter of time and applied science before computers can simulate election outcomes or the future price of stocks to any desired degree of accuracy. Computer scientist and polymath Stephen Wolfram lays out the argument in his ambitious twenty-year undertaking, A New Kind of Science:

让我能够为本书中描述的新型科学构建统一框架的关键思想是，正如任何系统的规则都可以被视为与程序相对应一样，它的行为也可以被视为与计算相对应^。18

The crucial idea that has allowed me to build a unified framework for the new kind of science that I describe in this book is that just as the rules for any system can be viewed as corresponding to a program, so also its behavior can be viewed as corresponding to a computation.¹⁸

沃尔夫勒姆的计算等价原理强烈宣称所有复杂系统从根本上来说都是计算的，并且，正如他在自己的工作与理论物理学和哲学等既定领域的联系中所暗示的那样，他相信计算主义提供了“一种真正可以找到（宇宙基本理论）的可能性” ^。19计算隐喻可以开启一种新的科学研究范式，这种观点对物理系统、社会行为和意识等的本质有着巨大的影响，在极端情况下，它成为那些寻求使用计算系统来建模和理解宇宙的人的一种超越意识形态。

Wolfram’s principle of computational equivalence makes the strong claim that all complex systems are fundamentally computational and, as he hints in the connections he draws between his work and established fields like theoretical physics and philosophy, he believes that computationalism offers “a serious possibility that [a fundamental theory for the universe] can actually be found.”¹⁹ This notion that the computational metaphor could unlock a new paradigm of scientific inquiry carries with it tremendous implications about the nature of physical systems, social behavior, and consciousness, among other things, and at its most extreme serves as an ideology of transcendence for those who seek to use computational systems to model and understand the universe.

海尔斯援引沃尔夫勒姆以及计算机科学家同行哈罗德·莫洛维茨和爱德华·弗雷德金的观点，追溯了一种基于复杂性科学的通用计算意识形态的兴起：如果宇宙是一台巨型计算机，那么开发用于文化问题的计算模型（例如评估贷款申请或建模意识）不仅高效，而且在智力上也是必要的。这些模型现在可能并不完美，但随着我们的使用，它们会不断改进，因为它们采用与它们模拟的系统相同的计算构建模块。从更深层次上讲，计算主义表明，我们对计算的知识将解答许多基本问题：计算成为物理科学、理论数学和文化问题的通用解决方案。对知识的追求变成了对计算的追求，一种建模的解释学。

Citing Wolfram and fellow computer scientists Harold Morowitz and Edward Fredkin, Hayles traces the emergence of an ideology of universal computation based on the science of complexity: if the universe is a giant computer, it is not only efficient but intellectually necessary to develop computational models for cultural problems like evaluating loan applications or modeling consciousness. The models may not be perfect now but they will improve as we use them, because they employ the same computational building blocks as the system they emulate. On a deeper level, computationalism suggests that our knowledge of computation will answer many fundamental questions: computation becomes a universal solvent for problems in the physical sciences, theoretical mathematics, and culture alike. The quest for knowledge becomes a quest for computation, a hermeneutics of modeling.

但模型当然总是压缩或简化现实。如果实用主义者对算法的定义基点是其基于对问题和解决方案的默契理解而具有的难以言喻的灵活性，那么这里的基点就是抽象的概念。计算主义的论证始于通用图灵机，这是数学家艾伦·图灵提出的一个令人惊叹的计算机愿景，它只需读取和写入标有1和0的无限磁带，并根据机器的当前状态向前或向后移动磁带，即可完成任何有限计算。仅使用这种简单的机制，就可以模拟任何类型的计算机，从计算曲线下面积的科学计算器到让马里奥在电视屏幕上移动的任天堂游戏机。换句话说，这建立了一个计算“天花板”，任何图灵计算机都可以模拟任何其他图灵计算机：指令执行速度可能更慢或更快，但在数学上是等价的。

But of course models always compress or shorthand reality. If the anchor point for the pragmatist’s definition of the algorithm is its indefinable flexibility based on tacit understanding about what counts as a problem and a solution, the anchor point here is the notion of abstraction. The argument for computationalism begins with the Universal Turing Machine, mathematician Alan Turing’s breathtaking vision of a computer that can complete any finite calculation simply by reading and writing to an infinite tape marked with 1s and 0s, moving the tape forward or backward based on the current state of the machine. Using just this simple mechanism one could emulate any kind of computer, from a scientific calculator finding the area under a curve to a Nintendo moving Mario across a television screen. In other words, this establishes a computational “ceiling” where any Turing computer can emulate any other: the instructions may proceed more slowly or quickly, but are mathematically equivalent.

通用图灵机是一个思想实验，旨在确定可计算范围：图灵和他的同事数学家阿隆佐·丘奇都在努力解决数学的边界问题。数学家大卫·希尔伯特提出了一个框架，称为判定问题 (Entscheidungsproblem)，问题是是否有可能预测某个程序何时或是否会停止，并在结束时给出或不给出答案。他们对希尔伯特的回应（现在称为丘奇-图灵论题）为理论家们定义了一种被广泛接受但最终无法证明的算法：使用自然数（或我们大多数人所知的整数）进行的计算，只有在通用图灵机能够完成的情况下才是“有效可计算的”（也就是说，只要有足够的时间和铅笔，人类就可以完成）。该论文使用这个非正式定义来统一三个不同的关于计算的严格数学论点（图灵机、丘奇的 lambda 演算和数学家库尔特·哥德尔的递归函数概念），将它们的具体数学主张转化为关于计算抽象极限的更一般的边界陈述。

The Universal Turing Machine is a thought experiment that determines the bounds of what is computable: Turing and his fellow mathematician Alonzo Church were both struggling with the boundary problems of mathematics. In one framing, posed by mathematician David Hilbert, known as the Entscheidungsproblem, the question is whether it’s possible to predict when or if a particular program will halt, ending its calculations with or without an answer. Their responses to Hilbert, now called the Church–Turing thesis, define algorithms for theorists in a way that is widely accepted but ultimately unprovable: a calculation with natural numbers, or what most of us know as whole numbers, is “effectively computable” (that is, given enough time and pencils, a human could do it) only if the Universal Turing Machine can do it. The thesis uses this informal definition to unite three different rigorous mathematical theses about computation (Turing machines, Church’s lambda calculus, and mathematician Kurt Gödel’s concept of recursive functions), translating their specific mathematical claims into a more general boundary statement about the limits of computational abstraction.

换言之，正如大卫·柏林斯基在其数学史著作《算法的出现》中所说，图灵、哥德尔和丘奇所努力探索的可计算性边界，也是对数理逻辑深层基础的探究。20^哥德尔证明，一个符号逻辑系统不可能在内部保持一致，也不可能仅使用系统内的语句就能证明，这让很多人感到沮丧。这样一个系统的真实性主张或验证，总是取决于一些外部的逻辑有效性假设或断言：就像乌龟一样一直往下沉。丘奇努力解决这个问题，并发展了 λ 演算，这是对抽象的精湛演示，在他之后的几十年里，它成为了众多编程语言的哲学基础。21^正如柏林斯基所说，图灵拥有“一种不可思议的、几乎万无一失的能力，可以筛选他那个时代的著作，并在筛选过程中辨别出比其他人所见事物简单得多的东西的轮廓。” ²²换句话说，他拥有抽象的天赋，而他在这方面最伟大的成就就是图灵机。

In another framing, as David Berlinski argues in his mathematical history The Advent of the Algorithm, the computability boundary that Turing, Gödel, and Church were wrestling with was also an investigation into the deep foundations of mathematical logic.²⁰ Gödel proved, to general dismay, that it was impossible for a symbolic logical system to be internally consistent and provable using only statements within the system. The truth claim or validation of such a system would always depend on some external presumption or assertion of logical validity: turtles all the way down. Church grappled with this problem and developed the lambda calculus, a masterful demonstration of abstraction that served as the philosophical foundation for numerous programming languages decades after his work.²¹ As Berlinski puts it, Turing had “an uncanny and almost unfailing ability to sift through the work of his time and in the sifting discern the outlines of something far simpler than the things that other men saw.”²² In other words, he possessed a genius for abstraction, and his greatest achievement in this regard was the Turing machine.

图灵的简单想象机是通用计算的优雅数学证明，但它也是一种原始算法，一个抽象生成器。丘奇与图灵工作之间的数学等价性很快表明，各种有效可计算性的证明（目前已有超过三十种）都指向某些基本的普遍真理。但每个抽象都有一个影子，在将某个想法提升到更高的思想层面的过程中，会留下一些模糊的背景和特殊性。图灵机留下了一个悬而未决的问题：“有效可计算”在物质现实中究竟意味着什么，在那里，我们告别了优雅和无限的磁带。随着丘奇-图灵论题从一个思想实验演变为计算主义的奠基信条（以及20世纪和21世纪计算革命的蓝图），它已经发展出一种引力，许多人感受到这种引力，促使他们按照宇宙的逻辑来组织宇宙。通用计算的概念在其核心中编码了一种直观的“有效”概念：可通过有限的步骤实现，并达到某种预期结果。因此，从一开始，算法就编码了一种特殊的抽象，即对答案渴望的抽象。这些计算中形成性证明的惊人清晰度和严谨性与该术语在计算机科学领域及其他领域的使用方式极其模糊形成了鲜明对比^。23

Turing’s simple imaginary machine is an elegant mathematical proof for universal computation, but it is also an ur-algorithm, an abstraction generator. The mathematical equivalence of Church and Turing’s work quickly suggested that varying proofs of effective computability (there are now over thirty) all gesture toward some fundamental universal truth. But every abstraction has a shadow, a puddled remainder of context and specificity left behind in the act of lifting some idea to a higher plane of thought. The Turing machine leaves open the question of what “effectively computable” might really mean in material reality, where we leave elegance and infinite tapes behind. As it has evolved from a thought experiment to a founding tenet of computationalism (and the blueprint for the computational revolution of the twentieth and twenty-first centuries), the Church–Turing thesis has developed a gravitational pull, a tug many feel to organize the universe according to its logic. The concept of universal computation encodes at its heart an intuitive notion of “effective”: achievable in a finite number of steps, and reaching some kind of desired result. From the beginning, then, algorithms have encoded a particular kind of abstraction, the abstraction of the desire for an answer. The spectacular clarity and rigor of these formative proofs in computation exists in stark contrast to the remarkably ill-defined way that the term is deployed in the field of computer science and elsewhere.²³

这种蕴含在有效性概念中的渴望在计算领域中通常被掩盖，但抽象的作用却备受推崇。通用图灵机提供了一个统一各种计算的概念平台：用于解决粒子物理学中一系列问题的算法可能突然在遗传学中变得有用；网络分析可以用来分析和比较书籍、商业网络和公交系统。抽象本身是丘奇-图灵论题——以及一般意义上的计算——赋予我们的最强大的工具之一，它使平台无关的软件以及我们所依赖的众多隐喻和视觉抽象（例如桌面用户界面）成为可能。

This desire encoded in the notion of effectiveness is typically obscured in the regime of computation, but the role of abstraction is celebrated. The Universal Turing Machine provides a conceptual platform for uniting all kinds of computing: algorithms for solving a set of problems in particle physics might suddenly be useful in genetics; network analysis can be deployed to analyze and compare books, business networks, and bus systems. Abstraction itself is one of the most powerful tools the Church–Turing thesis—and computation in general—gives us, enabling platform-agnostic software and the many metaphors and visual abstractions we depend on, like the desktop user interface.

抽象是 Wolfram 等人用来从特定计算系统攀登到通用计算概念的阶梯。许多复杂系统都表现出计算特性或看起来是可计算的。如果复杂系统本身就是计算图灵机，那么它们是等价的：天气系统、人类认知，以及最具挑衅性的宇宙本身。24^宇宙的宏大问题（宇宙的起源、时间和空间的关系）和文化的较小问题（票房收入、智能网络搜索、自然语言处理）是不可简化的，但也是可计算的：它们不是有简单答案的复杂问题，而是产生复杂答案的简单问题（或规则集）。这些假设打开了普遍数学的大门，普遍数学是哲学家戈特弗里德·威廉·莱布尼茨、勒内·笛卡尔等人预言的一种实现对自然世界完美理解的科学语言。²⁵这种完美的语言将通过其语法和词汇准确地描述宇宙，成为科学家有效描述和构成世界的一种新型理性魔法。

Abstraction is the ladder Wolfram et al. use to climb from particular computational systems to the notion of universal computation. Many complex systems demonstrate computational features or appear to be computable. If complex systems are themselves computational Turing Machines, they are therefore equivalent: weather systems, human cognition, and most provocatively the universe itself.²⁴ The grand problems of the cosmos (the origins thereof, the relationship of time and space) and the less grand problems of culture (box office returns, intelligent web searching, natural language processing) are irreducible but also calculable: they are not complicated problems with simple answers but rather simple problems (or rule-sets) that generate complicated answers. These assumptions open the door to a mathesis universalis, a language of science that the philosophers Gottfried Wilhelm Leibniz, René Descartes, and others presaged as a way to achieve perfect understanding of the natural world.²⁵ This perfect language would exactly describe the universe through its grammar and vocabulary, becoming a new kind of rational magic for scientists that would effectively describe and be the world.

有效可计算性在今天仍然是一个诱人而又模棱两可的术语，是实用主义和计算主义算法定义之间的一道断层线。我认为这是计算的第一次诱惑，根植于丘奇-图灵论题的核心。随着计算能力的增长，它的影响力不断扩大，与理性主义的根源相联系，逐渐成为一种更深层次、更浪漫的宇宙计算本体论神话。让世界有效可计算的愿望推动了计算机历史上许多开创性的时刻，从第一台弹道计算机在 20 世纪中叶的导弹防御系统中取代人类，到 Siri 和谷歌搜索栏。26^正是这种意识形态支撑了算法时代，它对人类知识和复杂系统地位的诱人主张构成了文化与文化机器关系的核心张力。

Effective computability continues to be an alluring, ambiguous term today, a fault line between the pragmatist and computationalist definition of algorithms. I think of this as computation’s first seduction, rooted at the heart of the Church–Turing thesis. It has expanded its sway with the growth of computing power, linking back to the tap root of rationalism, gradually becoming a deeper, more romantic mythos of a computational ontology for the universe. The desire to make the world effectively calculable drives many of the seminal moments of computer history, from the first ballistics computers replacing humans in mid-century missile defense to Siri and the Google search bar.²⁶ It is the ideology that underwrites the age of the algorithm, and its seductive claims about the status of human knowledge and complex systems in general form the central tension in the relationship between culture and culture machines.

为了理解有效可计算性的后果，我们需要遵循三个相互交织的线索，因为这一思想的含义在各个学科和文化领域中发挥作用：控制论、符号语言和技术认知。

To understand the consequences of effective computability, we need to follow three interwoven threads as the implications of this idea work themselves out across disciplines and cultural fields: cybernetics, symbolic language, and technical cognition.

主题 1：体现机器

Thread 1: Embodying the Machine

“有效可计算性”这一理念不仅影响着我们对人类在宇宙中地位的理解，也影响着我们如何理解生物、文化和社会系统。莱布尼茨关于“万能数学”（mathesis universalis）的设想极具吸引力，因为它承诺一套智力工具就能解开所有谜团，从量子力学到人脑内部的电路。二战后，一个致力于追求这一承诺的新领域应运而生，致力于将数学与物质性结合起来，探索计算与物理和社会科学之间的直接关联。在其鼎盛时期，控制论（该领域被称为控制论）引发了一场关于算法在物质文化中地位的持续争论——一场关于在物理和生物系统中实施数学思想，或声称数学思想体现在物理和生物系统中的政治争论。

“Effective computability” is an idea with consequences not just for our conception of humanity’s place in the universe but how we understand biological, cultural, and social systems. Leibniz’s vision of a mathesis universalis is seductive because it promises that a single set of intellectual tools can make all mysteries accessible, from quantum mechanics to the circuits inside the human brain. After World War II, a new field emerged to pursue that promise, struggling to align mathematics and materiality, seeking to map out direct correlations between computation and the physical and social sciences. In its heyday cybernetics, as the field was known, was a sustained intellectual argument about the place of algorithms in material culture—a debate about the politics of implementing mathematical ideas, or claiming to find them embodied, in physical and biological systems.

博学的数学家诺伯特·维纳于 1949 年出版了这门新学科的奠基性著作，称之为《控制论；动物与机器的控制与通信》。维纳称莱布尼茨为控制论的守护神：“莱布尼茨的哲学围绕着两个密切相关的概念——通用符号体系和推理演算。” ²⁷正如书名所示，20 世纪 40 年代和 50 年代的控制论的目标是定义和实现这两个理念：一个可以涵盖所有科学领域的知识体系，以及一种量化该系统内变化的方法。早期的控制论专家利用它们，试图在计算机科学、信息论、物理学等新兴领域之间建立一种综合体（事实上，维纳提名他的守护神莱布尼茨为最后一位“完全掌握他那个时代所有知识活动”的人）。²⁸从理智上讲，这种综合的载体是信息理论领域和不同个体与集体实体之间通信的有序特征；从务实上讲，这种综合的载体是机械和计算系统测量、调制和指导此类通信的日益增强的能力。

The polymathic mathematician Norbert Wiener published the founding text of this new discipline in 1949, calling it Cybernetics; or Control and Communication in the Animal and the Machine. Wiener names Leibniz the patron saint of cybernetics: “The philosophy of Leibniz centers about two closely related concepts—that of a universal symbolism and that of a calculus of reasoning.”²⁷ As the book’s title suggests, the aim of cybernetics in the 1940s and 1950s was to define and implement those two ideas: an intellectual system that could encompass all scientific fields, and a means of quantifying change within that system. Using them, the early cyberneticians sought to forge a synthesis between the nascent fields of computer science, information theory, physics, and many others (indeed, Wiener nominated his patron saint in part as the last man to have “full command of all the intellectual activity of his day”).²⁸ The vehicle for this synthesis was, intellectually, the field of information theory and the ordering features of communication between different individual and collective entities, and pragmatically, the growing power of mechanical and computational systems to measure, modulate, and direct such communications.

从哲学层面来看，维纳的控制论愿景取决于20世纪从确定性到概率的转变。29^爱因斯坦相对论和量子力学的进步表明，不确定性或非确定性是宇宙的根本，观察总是会影响被观察的系统。这标志着启蒙运动时期一种特定的理性主义理想的取代，这种理想认为宇宙由简单、万能的、可以发现和掌握的定律运行。相反，正如20世纪和21世纪数学物理日益复杂所揭示的那样，我们越仔细地观察一个物理系统，概率就变得越重要。放弃桌子——唯物主义哲学家们古老的支柱——舒适的坚固性，并用概率原子云取而代之，这令人不安。然而，只有用概率——更重要的是，用概率语言——我们才能开始描述我们的相对论宇宙。

On a philosophical level, Wiener’s vision of cybernetics depended on the transition from certainty to probability in the twentieth century.²⁹ The advances of Einsteinian relativity and quantum mechanics suggested that uncertainty, or indeterminacy, was fundamental to the cosmos and that observation always affected the system being observed. This marked the displacement of a particular rationalist ideal of the Enlightenment, the notion that the universe operated by simple, all-powerful laws that could be discovered and mastered. Instead, as the growing complexity of mathematical physics in the twentieth and twenty-first centuries has revealed, the closer we look at a physical system, the more important probability becomes. It is unsettling to abandon the comfortable solidity of a table, that ancient prop for philosophers of materialism, and replace it with a probabilistic cloud of atoms. And yet only with probability—more important, a language of probability—can we begin to describe our relativistic universe.

但更令人不安的是，信息论这一密切相关领域的核心论点是，概率既适用于信息，也适用于物质现实。数学家克劳德·香农将信息定义为不确定性、意外和不可预测的新数据，从而创建了一种可量化的通信测量方法。^香农的框架影响了信号处理、密码学和其他几个领域的数十年研究，但它对重要因素的极其有限的看法，对当代对计算知识的理解产生了重大影响。这种信息测量方法与人们对知识的普遍文化理解截然不同，尽管它在控制论中得到了广泛的表达，尤其是在维纳面向大众的著作《人之用》中。维纳正是在这里奠定了计算大教堂的基石之一：“有效地生活就是拥有充足的信息。因此，沟通和控制属于人类内心生活的本质，就像它们属于人类的社会生活一样。” ³¹从其有限的理论意义上讲，信息为理解任何有组织的系统提供了一个共同的标准；从其更广泛的公共意义上讲，它成为了计算主义的前沿，一种量化模式的方法，从而统一了复杂性的生物物理和数学形式。

But far more unsettling, and the central thesis of the closely allied field of information theory, is the notion that probability applies to information as much as to material reality. By framing information as uncertainty, as surprise, as unpredicted new data, mathematician Claude Shannon created a quantifiable measurement of communication.³⁰ Shannon’s framework has informed decades of work in signal processing, cryptography, and several other fields, but its starkly limited view of what counts has become a major influence in contemporary understandings of computational knowledge. This measurement of information is quite different from the common cultural understanding of knowledge, though it found popular expression in cybernetics, particularly in Wiener’s general audience book The Human Use of Human Beings. This is where Wiener lays one of the cornerstones for the cathedral of computation: “To live effectively is to live with adequate information. Thus, communication and control belong to the essence of man’s inner life, even as they belong to his life in society.”³¹ In its limited theoretical sense, information provided a common yardstick for understanding any kind of organized system; in its broader public sense, it became the leading edge of computationalism, a method for quantifying patterns and therefore uniting biophysical and mathematical forms of complexity.

正如维纳所说，信息对于控制论的关键价值在于决策。32通信和控制成为计算语言，通过它可以将生物系统、社会结构和物理学统一起来。正如海尔斯在《我们如何成为后人类》一书中所论证的^那样，早期的麦卡洛克-皮茨神经元（逻辑学家沃尔特·皮茨证明它在计算上等同于图灵机）等生物物理现实的理论模型使得控制论能够在范式和操作层面建立计算和生物过程之间的关联，并声称自己是信息学学者杰弗里·鲍克所说的“通用学科”。33^通过控制论，信息成为了“有效可计算性”扩展到广阔新领域的旗帜，首先呈现出沃尔夫勒姆等人后来追求的通用计算的诱人前景。³⁴早在《人之用处》一书中，维纳就普及了图灵机、神经网络和生物体学习之间的联系，而这项工作如今在谷歌子公司 DeepMind 过去几年宣布的一系列机器学习突破中变得令人震惊。

As Wiener’s quote suggests, the crucial value of information for cybernetics was in making decisions.³² Communication and control became the computational language through which biological systems, social structures, and physics could be united. As Hayles argues in How We Became Posthuman, theoretical models of biophysical reality like the early McCulloch–Pitts Neuron (which the logician Walter Pitts proved to be computationally equivalent to a Turing machine) allowed cybernetics to establish correlations between computational and biological processes at paradigmatic and operational levels and lay claim to being what informatics scholar Geoffrey Bowker calls a “universal discipline.”³³ Via cybernetics, information was the banner under which “effective computability” expanded to vast new territories, first presenting the tantalizing prospect that Wolfram and others would later reach for as universal computation.³⁴ As early as The Human Use of Human Beings, Wiener popularized these links between the Turing machine, neural networks, and learning in biological organisms, work that is now coming to startling life in the stream of machine learning breakthroughs announced by the Google subsidiary DeepMind over the past few years.

维纳由此攀登抽象的阶梯，将控制论定位为一种新的莱布尼茨式的普遍数学，能够统一各个领域。这一更高层次攀登的核心是体内平衡的概念，即系统如何响应反馈以保持其核心模式和身份。鸟儿在风向变化中保持高度，恒温器控制着房间的温度，以及古代神话​​代代相传，都是体内平衡发挥作用的例子。更具启发性的是，维纳提出，如果“有机体被视为信息，那么体内平衡可能与身份或生命本身是同一回事。有机体与混乱、解体和死亡相对立，就像信息与噪音相对立一样。” ³⁵这种论证思路演变成了哲学家温贝托·马图拉纳和弗朗西斯科·瓦雷拉在20世纪70年代提出的自创生理论，这是控制论的第二次浪潮，它将体内平衡的模式保存理论更充分地运用到了生物系统的语境中。将生物体描述为信息也暗示了相反的观点，即信息具有生存的意志，正如斯图尔特·布兰德的名言所说，“信息想要自由。” ³⁶

This is Wiener ascending the ladder of abstraction, positioning cybernetics as a new Liebnitzian mathesis universalis capable of uniting a variety of fields. Central to this upper ascent is the notion of homeostasis, or the way that a system responds to feedback to preserve its core patterns and identity. A bird maintaining altitude in changing winds, a thermostat controlling temperature in a room, and the repetition of ancient myths through the generations are all examples of homeostasis at work. More provocatively, Wiener suggests that homeostasis might be the same thing as identity or life itself, if “the organism is seen as message. Organism is opposed to chaos, to disintegration, to death, as message is to noise.”³⁵ This line of argument evolved into the theory of autopoiesis proposed by philosophers Humberto Maturana and Francisco Varela in the 1970s, the second wave of cybernetics which adapted the pattern-preservation of homeostasis more fully into the context of biological systems. Describing organisms as information also suggests the opposite, that information has a will to survive, that as Stewart Brand famously put it, “information wants to be free.”³⁶

如同尼尔·斯蒂芬森的可编程思维，如同那些试图模拟人脑的人工智能研究人员一样，这种将生物体视为信息的概念重新定义了生物学（以及人类），使其至少在理想情况下存在于有效可计算性的范围内。控制论和自创生催生了复杂性科学，并促使人们努力在模拟中模拟这些过程。例如，数学家约翰·康威的生命游戏就试图从简单的规则集精确地模拟这种信息的自发生成，或者说，看似鲜活或自我延续的模式。它也已被证明在数学上等同于图灵机，事实上，数学家保罗·伦德尔设计了一个生命游戏，并证明了其与图灵机等价（图1.1）。³⁷

Like Neal Stephenson’s programmable minds, like the artificial intelligence researchers who seek to model the human brain, this notion of the organism as message reframes biology (and the human) to exist at least aspirationally within the boundary of effective computability. Cybernetics and autopoiesis lead to complexity science and efforts to model these processes in simulation. Mathematician John Conway’s game of life, for example, seeks to model precisely this kind of spontaneous generation of information, or seemingly living or self-perpetuating patterns, from simple rule-sets. It, too, has been shown to be mathematically equivalent to a Turing machine, and indeed mathematician Paul Rendell designed a game of life that he proved to be Turing-equivalent (figure 1.1).³⁷

事实上，如果我们接受有机体作为信息的前提，接受信息模式作为生物生命的核心组织逻辑，我们就不可避免地会依赖计算作为探索该前提的框架。维纳从确定性转向概率的开场白，取代了但并未消除启蒙运动时期对普遍一致知识的旧目标。如今，这一抱负转向构建最佳模型，对现实复杂概率过程进行最精细的模拟。柏林斯基在解析微积分和计算微积分的区别中也观察到了同样的趋势，他指出，对难以处理的微分方程进行离散建模使我们能够更好地理解复杂系统的运作方式，但这总是以牺牲对事物在时间和数值上离散的近似视图为代价。38^对控制论的接受越来越多地意味着对社会、生物和物理系统的计算模拟作为研究的核心对象。

In fact, if we accept the premise of organism as message, of informational patterns as a central organizing logic for biological life, we inevitably come to depend on computation as a frame for exploring that premise. Wiener’s opening gambit of the turn from certainty to probability displaced but did not eliminate the old Enlightenment goals of universal, consilient knowledge. That ambition has now turned to building the best model, the finest simulation of reality’s complex probabilistic processes. Berlinski observed the same trend in the distinction between analytic and computational calculus, noting how the discrete modeling of intractable differential equations allows us to better understand how complex systems operate, but always at the expense of gaining a temporally and numerically discrete, approximated view of things.³⁸ The embrace of cybernetic theory has increasingly meant an embrace of computational simulations of social, biological, and physical systems as central objects of study.

海尔斯在《我们如何成为后人类》一书中仔细追溯了控制论中的这条线索，她认为，在梅西会议上，维纳和他的同事们敲定了控制论理论的愿景，也标志着一场通过抽象来抹去信息具身性的协同努力。在这些早期对话的记录、信件和其他档案材料中，她认为，控制论的综合抱负导致参与者在推进理论时回避了对反身性和具身性（尤其是人类具身性）的复杂性的思考。但是，正如海尔斯所说，“面对如此强大的梦想，想起信息的存在必须始终在媒介中实例化，可能会令人震惊。” ³⁹

Hayles traces this plumb line in cybernetics closely in How We Became Posthuman, arguing that the Macy Conferences, where Wiener and his collaborators hammered out the vision for a cybernetic theory, also marked a concerted effort to erase the embodied nature of information through abstraction. In the transcripts, letters, and other archival materials stemming from these early conversations, she argues that the synthesizing ambitions of cybernetics led participants to shy away from considerations of reflexivity and the complications of embodiment, especially human embodiment, as they advanced their theory. But, as Hayles puts it, “In the face of such a powerful dream, it can be a shock to remember that for information to exist, it must always be instantiated in a medium.”³⁹

虽然海尔斯对控制论的解读在构建“信息如何失去实体”的故事时，追求的是该领域在抽象阶梯上的修辞攀登，但 20 世纪 40 年代和 50 年代的控制论时刻还有另一面，它直接影响了硅谷的崛起以及人们普遍将计算系统理解为物质制品。我们也可以通过第二个关键的控制论术语“反馈”的概念，追溯维纳的抽象阶梯。正如海尔斯所指出的，反馈回路之所以引起维纳的兴趣，主要是因为它是一个通用的智力模型，可以用来理解如何在不同的系统中推广通信和控制。40^但反馈回路也是控制论应用的关键时刻，理论模型通过经验实验，或许更重要的是，通过演示进行了检验。

While Hayles’s reading of cybernetics pursues the field’s rhetorical ascent of the ladder of abstraction as she frames the story of “how information lost its body,” there is a second side to the cybernetic moment in the 1940s and 1950s, one that fed directly into the emergence of Silicon Valley and the popular understanding of computational systems as material artifacts. We can follow Wiener back down the ladder of abstraction, too, through a second crucial cybernetic term, the notion of “feedback.” The feedback loop, as Hayles notes, is of interest to Wiener primarily as a universal intellectual model for understanding how communication and control can be generalized across different systems.⁴⁰ But the feedback loop was also a crucial moment of implementation for cybernetics, where the theoretical model was tested through empirical experiments and, perhaps more important, demonstrations.

以维纳的“飞蛾”或“臭虫”为例，这是一台旨在演示与寻找或避开光线相关的反馈回路的机器。维纳与电气工程师杰里·威斯纳合作创造了这台机器，这是一个简单的机械装置，一个光电管朝右，另一个朝左，它们的输入控制一个“舵柄”机构，该机构会在车轮移动时瞄准车轮。该演示实现了其预期目的，即通过简单的反馈机制展示逼真的行为，从而为机械和生物物理控制机制的相似性以及控制论作为解释它们的模型的有效性提供了一个看似存在的证明。事实上，正如历史学家罗纳德·克莱恩所描述的，整个项目都是一个公关噱头，这个机器人的建造是由《生活》杂志资助的，该杂志计划刊登一篇关于控制论的文章。⁴¹维纳的演示机器预示了未来人机交互的奇观，就像早期硅谷偶像道格拉斯·恩格尔巴特的“所有演示之母”一样，它在 1968 年首次展示了功能性个人电脑体验的几个方面。

Consider Wiener’s “moth” or “bedbug,” a single machine designed to demonstrate a feedback loop related to seeking or avoiding light. Wiener worked with electrical engineer Jerry Wiesner to create the machine, a simple mechanical apparatus with one photocell facing to the right and another to the left, with their inputs directing a “tiller” mechanism that would aim the cart’s wheels as they moved. The demonstration achieved its intended purpose of showing lifelike behavior from a simple feedback mechanism, creating a seeming existence proof both of the similarity of mechanical and biophysical control mechanisms and of the efficacy of cybernetics as the model for explaining them. In fact, as historian Ronald Kline describes, the entire enterprise was a public relations stunt, the construction of the robot financed by Life magazine, which planned to run an article on cybernetics.⁴¹ Wiener’s demonstration machine presaged future spectacles of human–machine interaction like early Silicon Valley icon Douglas Engelbart’s “mother of all demos,” which first showcased several aspects of a functional personal computer experience in 1968.

控制论的理论抱负始终依赖于物质实现，这一事实挑战了一代又一代追求柏拉图式理想的人工智能研究人员，他们希望神经网络能够有效地模拟人类思维。42^克莱恩报道说，《生活》杂志从未刊登过维纳飞蛾的照片，因为一位编辑认为这台机器“通过模拟神经系统来展示人与机器的相似性，而不是展示计算机的人性特征，而这才是《生活》杂志的目标。” ⁴³最终，维纳制造了一个窃听器。飞蛾的物质背景不仅包括带轮子的正常反馈机制，还包括观察这一结构的文化视角。在实现过程中，机械反馈回路被智力回路所掩盖，即公共科学家和他在《生活》杂志的编辑之间的关系。事实证明，他们对维纳的论点（即反馈机制可以通过计算和机械建模）的兴趣远不及对在机器中寻找人性的兴趣。

The theoretical aspirations of cybernetics were always dependent on material implementation, a fact that has challenged generations of artificial intelligence researchers pursuing the platonic ideal of neural networks that effectively model the human mind.⁴² Kline reports that Life never ran photos of Wiener’s moth because an editor felt the machine “illustrated the analogy between humans and machines by modeling the nervous system, rather than showing the human characteristics of computers, which was Life’s objective.”⁴³ In the end, Wiener had built a bug. The material context of the moth included not just a functioning feedback mechanism on wheels but the cultural aperture through which that construct would be viewed. In implementation, the mechanical feedback loop was overshadowed by an intellectual one, the relationship between a public scientist and his editors at Life. As it turned out they were less interested in Wiener’s argument, that feedback mechanisms could be computationally and mechanically modeled, than they were in searching out the human in the machine.

主题二：魔法的隐喻

Thread 2: Metaphors for Magic

控制论最主要的尝试在于，它试图为沟通创造一种新的控制隐喻，将技术、生物和社会形式的知识融为一体。维纳飞蛾的故事揭示了这种方法的危险性：定义一个沟通的控制隐喻，进而定义知识的控制隐喻，需要深入研究语言本身如何塑造思想和现实。控制论关于统一的生物学和计算世界理解的愿景从未离开过我们，它不断出现在我们用来操纵和理解计算系统的技术和关键隐喻中。Chun 在《程序化愿景》中探讨了这种计算和生物学隐喻与代码持续交织的深层含义，展示了 DNA 研究与计算机编程之间的相互联系，以及这些隐喻如何揭示计算的解释性问题。对 Chun 来说，关键词是“软件”，她用这个词来涵盖我在算法语境中探讨的许多相同问题。

More than anything else, cybernetics was an attempt to create a new controlling metaphor for communication, one that integrated technological, biological, and social forms of knowledge. The story of Wiener’s moth illustrates the hazards of this approach: defining a controlling metaphor for communication, and by extension for knowledge, requires a deep examination of how language itself can shape both ideas and reality. The cybernetic vision of a unified biological and computational understanding of the world has never left us, continuing to reappear in the technical and critical metaphors we use to manipulate and understand computational systems. Chun explores the deeper implications of this persistent interlacing of computational and biological metaphors for code in Programmed Visions, demonstrating the interconnections of research into DNA and computer programming, and how those metaphors open up the interpretive problem of computation. For Chun the key term is “software,” a word she uses to encompass many of the same concerns I explore here in the context of the algorithm.

《程序化愿景》将图灵机具体化的可替代可计算性概念与定义我们许多计算体验的各种语言魔法直接联系起来：

Programmed Visions draws a direct link between the notion of fungible computability reified by the Turing machine and the kinds of linguistic magic that have come to define so many of our computational experiences:

软件的独特之处在于，它作为隐喻本身的隐喻。作为一种通用的模仿者/机器，它蕴含着一种普遍可替代性的逻辑；一种秩序化、创造性和活力化的无序化的逻辑。约瑟夫·魏森鲍姆认为，计算机已经成为“有效程序”的隐喻，即任何可以通过规定步骤解决的问题，例如基因表达和文书工作^。44

Software is unique in its status as metaphor for metaphor itself. As a universal imitator/machine, it encapsulates a logic of general substitutability; a logic of ordering and creative, animating disordering. Joseph Weizenbaum has argued that computers have become metaphors for “effective procedures,” that is, for anything that can be solved in a prescribed number of steps, such as gene expression and clerical work.⁴⁴

Chun 认为，在“普遍替代逻辑”的指导下，软件已成为一种物，体现了魔法的核心功能——以影响世界的方式操纵符号。这种基本的炼金术，即神秘的源码可互换性，强化了将图灵机解读为一种原始算法的解读，这种算法八十年来一直在其“用户”的头脑中大量产生有效的可计算性抽象。软件已成为的“物”，是算法的文化形象：有效程序的实例化隐喻。Chun 认为，软件就像博格斯特的计算大教堂，“一个强有力的隐喻，涵盖了我们认为看不见却产生看得见影响的一切事物，从遗传学到市场这只看不见的手，从意识形态到文化。” ⁴⁵就像十字架或标志着周日弥撒的钟楼一样，软件无处不在，即使它显而易见，也充满神秘，它以熟悉的形式出现，而这些形式只是其在幕后所做实际工作的象征性代表。

With the “logic of general substitutability,” software has become a thing, Chun argues, embodying the central function of magic—the manipulation of symbols in ways that impact the world. This fundamental alchemy, the mysterious fungibility of sourcery, reinforces a reading of the Turing machine as an ur-algorithm that has been churning out effective computability abstractions in the minds of its “users” for eighty years. The “thing” that software has become is the cultural figure of the algorithm: instantiated metaphors for effective procedures. Software is like Bogost’s cathedral of computation, Chun argues, “a powerful metaphor for everything we believe is invisible yet generates visible effects, from genetics to the invisible hand of the market, from ideology to culture.”⁴⁵ Like the crucifix or a bell-tower signaling Sunday mass, software is ubiquitous and mysterious even when it is obvious, manifesting in familiar forms that are only symbolic representations of the real work it does behind the scenes.

软件作为隐喻的隐喻的优雅表述，加上 Chun 引用了魏森鲍姆（Weizenbaum）——这位麻省理工学院的计算机科学家在 20 世纪 60 年代创造了一位令人震惊的成功算法心理治疗师 ELIZA——将控制论与魔法联系在一起，因为计算机本身已成为有效可计算性空间的隐喻。算法并非物质与符号秩序相互竞争的空间，而是一个神奇的、炼金术般的境界，它们在其中以富有成效的不确定性运作。算法跨越了代码与实现、软件与经验之间的鸿沟。

The elegant formulation of software as a metaphor for metaphor, paired with Chun’s quotation of Weizenbaum—the MIT computer scientist who created an alarmingly successful algorithmic psychotherapist called ELIZA in the 1960s—draws together cybernetics and magic through the notion that computers themselves have become metaphors for the space of effective computability. The algorithm is not a space where the material and symbolic orders are contested, but rather a magical or alchemical realm where they operate in productive indeterminacy. Algorithms span the gap between code and implementation, between software and experience.

从这个角度来看，计算是一种通用溶剂，正是因为它既是隐喻又是机器。就像维纳的机器飞蛾一样，所实施的算法一方面是一种智力姿态（“你好，世界！”），一种宣传噱头，另一方面是一个功能系统，在其构造中嵌入了关于感知、决策和沟通的物质假设。例如，想想不起眼的进度条。当一款新软件显示一个据称以图表形式显示安装速度的指标时，该代码很可能有点神奇（进度条的状态与幕后实际进行的工作几乎没有关系）。但对于用户来说，那熟悉的缓慢移动的进度条也是一种功能现实，因为无论映射的“进度”多么虚构，在进度条达到 100% 之前什么也不会发生——幻觉决定现实。进度条的算法不仅取决于生成它的代码，还取决于等待本身的文化计算、用户寻求系统反馈，以及在等待阶段向用户展示其他消息、娱乐或广告的机会（越来越被利用）。

In this light, computation is a universal solvent precisely because it is both metaphor and machine. Like Wiener’s robotic moth, the implemented algorithm is on the one hand an intellectual gesture (“Hello, world!”), a publicity stunt, and on the other a functioning system that embeds material assumptions about perception, decision-making, and communication in its construction. For example, think of the humble progress bar. When a new piece of software presents an indicator allegedly graphing the pace of installation, that code might well be a bit of magic (the status of the bar holding little relation to the actual work going on behind the scenes). But that familiar inching bar is also a functional reality for the user because no matter how fictitious the “progress” being mapped, nothing else is going to happen until the bar hits 100 percent—the illusion dictates reality. The algorithm of the progress bar depends not only on the code generating it but the cultural calculus of waiting itself, on a user seeking feedback from the system, and on the opportunity—increasingly capitalized on—to show that user other messages, entertainments, or advertising during the waiting phase.

正如我们通常不假思索地接受进度条所表明的那样，我们已经准备好在多个层面上接受这些神奇的计算。我们相信代码的力量，它是一组连接可见与不可见的神奇符号，与我们悠久的逻各斯文化传统相呼应，或者说，语言是秩序和理性的基础系统，它的力量是一种源泉。我们相信控制论的优雅抽象，以及最终的计算宇宙——算法以文化上可读的方式体现和再现现实的数学基础。这就是算法是文化机器的含义：它在有效可计算性的反身性障碍之内和之外运行，在生产文化对象、过程和体验的同时也在宏观社会层面上生产文化。

As our generally unthinking acceptance of the progress bar demonstrates, we are primed to accept these magical calculations on multiple levels. We believe in the power of code as a set of magical symbols linking the invisible and visible, echoing our long cultural tradition of logos, or language as an underlying system of order and reason, and its power as a kind of sourcery. We believe in the elegant abstractions of cybernetics and, ultimately, the computational universe—that algorithms embody and reproduce the mathematical substrate of reality in culturally readable ways. This is what it means to say that an algorithm is a culture machine: it operates both within and beyond the reflexive barrier of effective computability, producing culture at a macro-social level at the same time as it produces cultural objects, processes, and experiences.

然而，我们与这些文化机器越来越熟悉、越来越紧密地联系在一起，并不意味着我们能够更深入地理解它们魔力的本质。魏森鲍姆认为，即使是那些最接近这些奥秘的人，那些直接用代码实现算法的程序员和开发者，

Just because we are growing more familiar, more intimately entangled with these culture machines, however, does not mean we understand the nature of their magic any more deeply. Weizenbaum argues that even for those closest to the mysteries, the programmers and developers who directly implement algorithms in code,

工具理性把文字变成了一种被黑魔法包围的迷信。只有魔法师才拥有入门者的权利。只有他们才能说出文字的含义。他们玩弄文字，欺骗我们^。46

instrumental reason has made out of words a fetish surrounded by black magic. And only the magicians have the rights of the initiated. Only they can say what words mean. And they play with words and they deceive us.⁴⁶

Chun 扩展了 Weizenbaum 对“拜物教”的论述，认为对源代码的崇拜只会增强人们对计算以及那些运用这种力量的程序员的神秘感知。但她随后指出，我们可以通过某种渠道来尝试重建这种机制的隐形运作。

Chun extends Weizenbaum’s reference to fetish, arguing that the fetish of source code has only increased the perceived occult power of computation and the programmers who wield it. But, as she points out soon afterward, there is a channel through which we attempt to reconstruct the invisible working of the mechanism.

存在一种算法，一种代码所意图表达的含义（因此在某种程度上是可知的），这一事实有时会构建我们与程序的体验。当我们玩游戏时，我们可能会尝试对其算法进行逆向工程，或者至少将其操作与其编程联系起来，这就是为什么所有设计书籍都警告不要使用巧合或随机映射，因为它可能会诱发用户的偏执。也就是说，由于界面是经过编程的，大多数用户会将巧合视为有意义的。对用户来说，就像偏执型精神分裂症患者一样，它总是有意义的：无论用户是否了解其含义，他/她都知道它与他或她息息相关^。47

The fact that there is an algorithm, a meaning intended by code (and thus in some way knowable), sometimes structures our experience with programs. When we play a game, we arguably try to reverse engineer its algorithm or at the very least link its actions to its programming, which is why all design books warn against coincidence or random mapping, since it can induce paranoia in its users. That is, because an interface is programmed, most users treat coincidence as meaningful. To the user, as with the paranoid schizophrenic, there is always meaning: whether or not the user knows the meaning, s/he knows that it regards him or her.⁴⁷

这是一场以游戏代替世界的知识探索。Chun 揭示了另一种形式的魔法——在计算的信息体验背后插入目的或意义。当我们试图洞察界面的奥秘时，我们通常对体验如何运作（位存储在哪里，像素如何渲染）更感兴趣，而不是为什么。如果软件是隐喻的隐喻，那么算法就成为翻译的机制：通过棱镜或工具，有效可计算性的永恒可替代空间在特定的程序、界面或用户体验中被聚焦和实例化。带着我们的幻想性错觉，带着我们对有意义模式的拼命追寻，我们试图透过算法来瞥见它背后的计算。

This is a quest for knowledge where the game substitutes for the world. Chun reveals another form of magic—an interpolation of purpose or meaning behind the informational experiences of computation. When we attempt to penetrate the mysteries of the interface, we are usually far less interested in how the experience works (where the bits are stored, how the pixels are rendered) than we are in why. If software is a metaphor for metaphors, the algorithm becomes the mechanism of translation: the prism or instrument by which the eternally fungible space of effective computability is focalized and instantiated in a particular program, interface, or user experience. With our apophenia, our desperate hunt for meaningful patterns, we try to peer through the algorithm to catch a glimpse of the computation behind it.

这最纯粹地体现了语言本身作为一种有效传递意义和经验的系统，跨越了思维​​鸿沟的根本魔力。作为跨越计算空间和文化空间的工具，算法也充当着一座桥梁，有时可以实现双向交流：意义，或者至少是意义的承诺，以及一条解读的途径。在屏幕背后寻找这种意义，正是柏林斯基所说的“寻找异域智慧”的过程。他对此的追求最终得出了一个出人意料的结论：算法智能或许可以支持宇宙智能设计的论证。48^这座桥似乎有些过头，但这一论证强调了大教堂与计算之间的深层联系。从代码的模式、游戏和政治中抽象出来，去发现它们背后的意义，是算法阅读的根本步骤，我们将在下文中继续讨论。因此，Chun 的拜物教和柏林斯基的智能设计有一些共同之处：他们都认为计算的核心魔力依赖于对意义和理解的探索，而这种探索又依赖于人类特定的认知和识别形式。

This is the purest expression of the fundamental magic of language itself as a system for effectively transmitting meaning and experience across the gulf between minds. As the instrument for spanning computational and cultural spaces, the algorithm also serves as a bridge that can, at times, allow traffic in both directions: meaning, or at least the promise of meaning, and an avenue for interpretation. The search for that meaning behind the screen is what Berlinski calls the search for “intelligence on alien shores,” an idea he pursues to the unexpected conclusion that algorithmic intelligence might support an argument for the universe’s intelligent design.⁴⁸ This is a bridge too far, but the argument underscores the deep links between the cathedral and computation. Abstraction from the patterns, games, and politics of code to the discovery of meanings behind them is the essential move of algorithmic reading, which we return to below. Chun’s fetishism and Berlinski’s intelligent design have something in common, then: an argument that the central magic of computation depends on a quest for meaning, for understanding, that depends on particular forms of human cognition and recognition.

主题 3：如何思考算法

Thread 3: How to Think about Algorithms

追寻具身计算和语言力量的线索，直至得出结论，我们最终会遇到最具挑战性，或许也是最有趣的问题：心智本身如何应对有效的可计算性。我们对复杂性的迷恋以及对模式意义的探索，构成了我们关于如何在认知层面上与机器和工具互动的一些根深蒂固的迷思的根源。正因如此，像《雪崩》这样的故事才如此引人入胜：它们触及了一种根深蒂固的渴望，即融合内在和外在的秩序与意义结构。人类心智的通用操作系统蕴含着各种各样的含义，它关乎信息如何转化为我们具身物理自我内部和外部的知识——这些知识，如同控制论中的知识一样，可以在计算信息系统和生物信息系统之间精准地转化。计算与认知之间这种理想化联系的变体是双向的，从《黑客帝国》中即时下载功夫知识，到奇点倡导者所提出的将意识上传到数字计算机的愿景。然而，这些极端情况太过奇妙，以至于它们有时会掩盖算法已经改变我们思维方式的多种方式。

Following the threads of embodied computation and the power of language to their conclusions brings us to the most challenging, and perhaps most interesting, question of all: how the mind itself grapples with effective computability. Our fascination with complexity and the search for meaning in patterns underlies some of our most entrenched myths about how we relate to our machines and tools on a cognitive level. This is what makes stories like Snow Crash so compelling: they address a deep-seated desire to fuse internal and external structures of order and meaning. A universal operating system for the human mind carries with it all sorts of implications for how information can be transmuted into knowledge both within and beyond our embodied physical selves—knowledge that, as in cybernetics, translates precisely between computational and biological information systems. Variations on this idealized link between computation and cognition run in both directions, from instantly downloading the knowledge of kung fu in The Matrix to the vision advanced by advocates of the singularity for uploading consciousness onto a digital computer. Those extremes are so fantastic, however, that they sometimes obscure the many ways that algorithms are already changing how we think.

技术与意识的接口不仅仅是科幻小说中的东西。我们使用各种各样的工具来增强或改变我们对世界的心理体验。这些可以是算法系统，就像智能​​手机上那些复杂的通知和提醒系统，它们可以鼓励用户多锻炼或准时参加即将到来的会议。或者它们可以非常简单，就像盲人使用拐杖导航一样。事实上，正如柏拉图在《斐德罗篇》中所说的那样，写作本身就是这样一种技术，他担心写作会削弱那些依赖写作而不是自身记忆力和理解力的人的心智能力。49^从此，人类开始了将我们的思想外包给机器的悠久历史：将我们的思想和记忆托付给石头、纸莎草、平版印刷品、蜡盘、照相底片、硬盘和网络存储服务。人类记忆在技术维度上的扩展，使得我们今天能够“记住”比以往任何时候都多的东西，尽管它改变了我们进行某些生物记忆的能力，鼓励我们关注如何获取信息，而不是信息本身^。50

Technological interfaces with consciousness are not only the stuff of science fiction. We use all sorts of tools to enhance or modify our mental experience of the world. These can be algorithmic systems, like those sophisticated notifications and nudges smartphones can deploy to encourage their owners to exercise more or depart on time for an upcoming meeting. Or they can be very simple, like a blind man using a cane to navigate. Indeed, as Plato famously argued in the Phaedrus, writing itself is such a technology, one he feared would diminish the mental faculties of those who depended on it instead of their own powers of memory and understanding.⁴⁹ Thus begins the long history of humanity outsourcing our minds to machines: entrusting our thoughts and memories to stone, papyrus, lithograph, wax platter, photographic negative, hard drive, and networked storage service. The extension of human memory in technological dimensions allows us to “remember” far more today than we ever did, even as it alters our capacity to perform certain kinds of biological recollection by encouraging us to focus on how we might access information rather than the information itself.⁵⁰

认知哲学家安迪·克拉克（Andy Clark）将其称为“延展心智”，这是一种认知框架，它涵盖了认知从意识大脑向身体以及我们周围的社会和技术环境的多种方式。51^随着我们与算法系统的互动和联系日益紧密，记忆、决策甚至欲望对外部资源的“耦合”或依赖也会变得更加强烈。52正如克拉克在《天生的机器人》（Natural-Born Cyborgs）^中所写，

The philosopher of cognition Andy Clark has called this the “extended mind,” a framing of cognition that accommodates the many ways in which it spills out of the conscious brain into the body and our surrounding social and technical environments.⁵¹ As we grow more engaged and connected to algorithmic systems, the “coupling” or dependency on external resources for memory, decision-making, and even desire can become stronger.⁵² As Clark writes in Natural-Born Cyborgs,

人类的思维和理性诞生于物质大脑、物质身体以及复杂的文化和技术环境之间的循环互动。我们创造了这些支持性环境，但它们也创造了我们^。53

Human thought and reason is born out of looping interactions between material brains, material bodies, and complex cultural and technological environments. We create these supportive environments, but they create us too.⁵³

很大程度上，正是因为这种技术扩展已经深深植根于人类文化，人机融合的可能性才如此令人担忧。柏拉图担忧的当代版本与斯蒂芬森的“nam-shub”（即人类思维与计算过程可以被证明是等价的）源于同一源头。正如魏泽鲍姆所言，“人类思维是否完全可计算”这一问题，在科学主义和对现代技术系统的盲目崇拜的幌子下，导致了“现代科学所产生的精神宇宙学[被]逻辑必然性的萌芽所感染” ^。54这与维纳从确定性到概率的转变相呼应，“它们将真理转化为可证明性”。55^魏泽鲍姆认为，我们痴迷于掌握有效可计算性范围内的一切，这不仅蒙蔽了我们，使我们看不到边界之外的事物，也削弱了我们理解或讨论计算在其领域内缺陷的能力。 “对理性-逻辑等式的信仰已经腐蚀了语言本身的预言能力。我们可以计算，但我们正在迅速忘记如何表达什么值得计算以及为什么。” ⁵⁶我们混淆了知识和意义、过程和目的，用最初启蒙运动对知识追求的目的论，即对量化和可计算性的追求，取代了后者。

It is in large part because this kind of technological extension is already so deeply embedded in human culture that the possible integration of human and computer has been so fraught. The contemporary version of Plato’s concern springs from the same source as Stephenson’s nam-shub—the idea that human thought and computational processes can be shown to be equivalent. The question, as Weizenbaum puts it, of “whether or not human thought is entirely computable” has led, in the guise of scientism and the blind worship of modern technical systems, to the “spiritual cosmologies engendered by modern science [becoming] infected with the germ of logical necessity.”⁵⁴ Echoing Wiener’s move from certainty to probability, “they convert truth to provability.”⁵⁵ Weizenbaum suggests that our obsession with mastering everything that falls within the boundaries of effective computability has not only blinded us to what lies beyond that frontier, but also weakened our capacity to understand or debate the failings of computation within its domain. “Belief in the rationality-logicality equation has corroded the prophetic power of language itself. We can count, but we are rapidly forgetting how to say what is worth counting and why.”⁵⁶ We confuse knowledge and meaning, process and purpose, by substituting the teleology of the original Enlightenment quest for knowledge with that secondary substitute, the quest for quantification and calculability.

延伸这一思路，魏泽鲍姆最困扰的并非计算机直接模拟人类思维或用硅片建模的思维，而是计算思维对人类自身的腐蚀性影响。“如今，语言已沦为另一种工具，所有艺术家和作家无法用计算机可理解的语言诠释的概念、思想和图像都失去了它们的功能和效力。” ⁵⁷戈伦比亚（Golumbia）认为，计算主义是一种政治上危险的意识形态（我们将在下文中讨论），但魏泽鲍姆在此的根本关注点在于语言想象力和文字的“预言力”。计算的引力，即拉力，促使我们规范和整理我们的语言，在思想之间建立持久且同构的关系，并根据关系数据库的表示形式定制我们的思维。在《雪崩》中，黑客最容易受到 nam-shub 的影响，因为计算思维已经重新组织了他们的思想：“你的神经在使用时会产生新的连接——轴突分裂并挤入分裂的神经胶质细胞之间——你的生物软件会自我修改——软件会成为硬件的一部分。” ⁵⁸斯蒂芬森在这里描述的过程是自动化，即重新调整心理能力以内化和稳态管理驾驶汽车等复杂任务。而且，正如媒体记者尼古拉斯·卡尔在《玻璃笼子》中指出的那样，我们在学习阅读时都亲身体验过自动化，逐渐将语法和拼写规则内化，直到对书面符号的解释变得毫不费力。59^正如柏拉图所担心的那样，我们与书面文字技术的互动不仅改变了思维媒介，将其扩展到外部纸张、卷轴和其他物质东西，而且还改变了思维方式。

Extending this line of thinking, what troubles Weizenbaum the most is not the vision of computers directly emulating human thought, or minds modeled in silicon, but rather the corrosive impact of computational thinking on the human self. “Now that language has become merely another tool, all concepts, ideas, images that artists and writers cannot paraphrase into computer-comprehensible language have lost their function and their potency.”⁵⁷ Golumbia takes up this polemical argument against computationalism as a politically dangerous ideology (to which we return below), but Weizenbaum’s fundamental concern here is about linguistic imagination and the “prophetic power” of words. The gravity, the pull of computation, encourages us to discipline and order our language, to establish persistent and isomorphic relationships between ideas, to tailor our thinking for representation in relational databases. In Snow Crash the hackers were most susceptible to the nam-shub because computational thinking had already reordered their minds: “Your nerves grow new connections as you use them—the axons split and push their way between the dividing glial cells—your bioware self-modifies—the software becomes part of the hardware.”⁵⁸ The process Stephenson describes here is automatization, the realignment of mental faculties to internalize and homeostatically manage a complex task like driving a car. And, as media journalist Nicholas Carr points out in The Glass Cage, we all experienced automatization first-hand when we learned to read, gradually internalizing the rules of grammar and spelling until the interpretation of written symbols was largely effortless.⁵⁹ Just as Plato feared, our interaction with the technology of the written word not only changed the medium of thought, extending it to external papers, scrolls and other material stuff, but it also changed the mode of thought.

我们如何看待算法是一个将符号语言、可计算性和大脑可塑性联系起来的问题。我们或许会认为，读写能力的内化是一种重新编程，一种我们一代又一代辛勤灌输给孩子的“nam-shub”（知识的积累）。60^我们或许会将麦卡洛克-皮茨神经元的现代实例付诸实践，并声称维纳的控制论可以计算地描述大脑神经网络的重构。这些都可以用来论证阅读是编码在读写能力者头脑中的一种心理算法。我的兴趣不在于探究这一概念，而在于引我们到这里的哲学基础：语言本身，尤其是书面语言，是人类思维的原始“外部”，是第一台跨越时空处理文化的机器。

How we think about algorithms is a question that links symbolic language, computability, and brain plasticity. We might argue that the internalization of literacy is a kind of reprogramming, a nam-shub that we laboriously inculcate in children with every passing generation.⁶⁰ We might press the modern instantiation of the McCulloch–Pitts neuron into service, claiming with Wiener’s cybernetics that the reconfiguring of neural networks in the brain can be computationally described. These would be pathways to arguing that reading is a mental algorithm encoded in literate minds. I am less interested in pursuing this notion than I am in the philosophical underpinnings that bring us here: the way that language itself, particularly written language, serves as the original “outside” to human thought, the first machine for processing culture across time and space.

语言作为一种认知技术的角色问题意义深远，它将丘奇-图灵论题的符号逻辑基础与魔法和文化建构现实的问题联系在一起。事实上，我们将语言视为人类与技术关系的一个特例，正是因为它在建构我们所感知的世界方面发挥着本体论的作用。安迪·克拉克为这一论题提供了一些令人信服的实证证据，他认为语言是“一种种子技术，它帮助设计师环境创造的整个过程得以实现”。61^语言作为一种自我系统，其技术功能不仅支撑着沟通，也支撑着“精确的数字推理”，而脑部扫描显示，精确的数字推理依赖于语言中枢^。62

The question of language’s role as a technology of cognition is a deep one, linking the Church-Turing thesis’s foundations in symbolic logic to the question of magic and culturally constructed reality. Indeed, we perceive language as a special case of the relationship between humanity and technology precisely because it plays an ontological role in constructing the world as we perceive it. Andy Clark offers several compelling pieces of empirical evidence for this thesis, arguing that language is “the kind of seed-technology that helped the whole process of designer-environment creation get off the ground.”⁶¹ The technological function of language as a system of me underpins not just communication but “precise numerical reasoning,” which brain scans have revealed depends on language centers.⁶²

魏森鲍姆在想象的背景下也提出了同样的观点：

Weizenbaum argues the same point in the context of imagination:

正是在个人所创造的智识与社会世界中，他预演并演练了无数关于世界原本可能是什么样子以及未来可能变成什么样子的戏剧性场景。这个世界是他主观性的宝库。……如果不先想象自己能够创造，人类就无法创造任何东西^。63

It is within the intellectual and social world he himself creates that the individual prehearses and rehearses countless dramatic enactments of how the world might have been and what it might become. That world is the repository of his subjectivity. … Man can create little without first imagining that he can create it.⁶³

工具和流程体现了这些实践，从史前第一把石斧，到如今搜索引擎呈现的信息宇宙模型。或者用Hayles的话来说，认知

Tools and processes are the me that embody these enactments, from the first prehistoric stone ax to the model of the informational universe that our search engines present us today. Or as Hayles puts it, cognition

延伸到技术环境中，将内部和外部的界限消解为流动的组合，将技术制品纳入人类认知系统，不仅作为隐喻，而且作为日常思想和行动的有效部分^。64

reaches out into the techno-environment, dissolving the boundary between inside and outside into fluid assemblages that incorporate technical artifacts into the human cognitive system, not just as metaphors but as working parts of everyday thoughts and actions.⁶⁴

第一个工具，即原始过程，是语言的主体间文化机器。

And the first tool, the ur-process, is the intersubjective culture machine of language.

将语言视为工具，也使我们能够将其他工具视为语言陈述，视为包含概念、语法和动词的“nam-shub”。正如魏森鲍姆雄辩地指出的那样，它们“本身就是蕴含着象征意义的符号”：“工具也是其自身再生产的一个模型，也是重演其所象征技能的脚本。” ⁶⁵这种思路与技术哲学家吉尔伯特·西蒙东关于技术的思考高度呼应，西蒙东探讨的是技术对象如何确立自身身份，并构成反映多种相互竞争的社会技术力量之间紧张关系的整体。海尔斯将西蒙东的整体概念与奈杰尔·斯里夫特的观点交织在一起，后者认为我们正在日益自动化技术系统，不再将其视为塑造我们世界的力量，而只是在潜意识层面接受它们的功能性和设计准则。⁶⁶这正是魏森鲍姆在二十年前警告的结果：轻易地从理性滑向对逻辑性的依赖，并且越来越多地依赖于对现实的计算近似。

To think of language as a tool also allows us to begin seeing our other tools as linguistic statements too, as nam-shubs that contain concepts, grammars, verbs. They are, as Weizenbaum eloquently puts it, “pregnant symbols in themselves”: “a tool is also a model for its own reproduction and a script for the reenactment of the skill it symbolizes.”⁶⁵ This line of thinking closely echoes the philosopher of technology Gilbert Simondon in his thinking about technics, or the ways in which technical objects can establish their own identities and constitute ensembles that reflect the tensions between multiple competing sociotechnical forces. Hayles weaves Simondon’s notion of ensembles together with Nigel Thrift’s argument that we are increasingly automatizing technological systems, ceasing to perceive them as forces that shape our world and simply accepting their functionality and design imperatives on a subconscious level.⁶⁶ This is precisely the outcome that Weizenbaum cautioned against twenty years avant la lettre: the easy slide from rationality to a dependency on logicality and, increasingly, computational approximations of reality.

在此阶段，关于技术本质及其后果的哲学思考思路，其具体关联性在我们讨论算法时逐渐显现。从柏拉图到西蒙东，关于我们智力对外部技术组合的依赖性的争论，与20世纪初关于数学可计算性和逻辑一致性的争论有着典型的相似性，最终催生了丘奇-图灵论题。让我来解释一下：数学家们开启有效可计算性之路，首先要探究符号语言的局限性。这是对数学思维基础及其边界的探究，其基础是认识到数学语言本身是数学思维机制的重要组成部分。这只是克拉克“延展心智”假说几十年后引发的更广泛的认知与心智争论的一个缩影——它考察了认知与认知工具之间的关系，并以数学真理性和可证明性为依据。

At this stage the specific relevance of this line of philosophical thinking about the nature and consequences of technics swings into view for our discussion of algorithms. The debate carried on from Plato to Simondon regarding our intellectual dependencies on external technical assemblages is paradigmatically similar to the debate over mathematical computability and logical consistency that raged in the early twentieth century, ultimately producing the Church–Turing thesis. Let me explain: mathematicians launched on the pathway to effective computability by first asking what the limits of symbolic languages were. This was an investigation of the foundations as well as the boundaries of mathematical thought based on the recognition that the languages of mathematics were themselves an essential part of the machinery of that thought. This was a limited instance of the broader cognition and mind debates Clark’s “extended mind” hypothesis sparked decades later—an examination of the relationship between cognition and the tools of cognition, grounded here in terms of mathematical truth and provability.

这场辩论关乎我们对数学语言依赖的本质，以及语言的选择（不同符号系统的可供性）如何阻碍我们获取其他理解方式。哥德尔不完备定理明确地否定了一个根本的、存在性的问题（是否存在一种合乎逻辑且完备的数学语言？）。他证明了，任何一种其语句可有效计算的数学语言，都无法既证明所有关于自然数的真语句，又能保持逻辑一致性。或许更糟糕的是，任何这样的系统都无法证明自身的一致性——为了证明这一点，我们必须始终超越这种语言的界限。这个定理是一个令人震惊的结果，它解决了几代人以来关于算术基本真理和数学基础的争论，旨在为迅速扩展的数学结构寻找哲学权威的基石。

It was a debate about the nature of our dependence on mathematical language and the ways that choices of language, the affordances of different symbolic systems, could foreclose access to other means of understanding. Gödel’s incompleteness theorem definitively answered a fundamental, existential question (is there a logical and complete mathematical language?) with a firm negative. He demonstrated that no mathematical language whose statements are effectively calculable can both prove all true statements about natural numbers and remain logically consistent. Perhaps more damning, no such system can demonstrate its own consistency—one must always reach outside the boundaries of such a language in order to prove it. This theorem was a startling result, addressing generations of debate over the fundamental truth of arithmetic and the foundations of mathematics, an effort to find a bedrock of philosophical authority to support the rapidly expanding structures of mathematics.

丘奇-图灵论题的发展和有效可计算性基础通过图灵机创造了一种新的语言机制。这个概念对象，这个抽象引擎，开辟了计算操作空间，并发挥了清晰表达我们对有限符号系统依赖性质的基本功能。多亏了这些证明，我们才知道界限在哪里。丘奇、图灵和其他人提供的有效可计算性的数学证明创造了一种新的确定性和一种思考的新隐喻——分别是可计算性的通用数学上限和图灵机。但是，它们也在有效可计算性的边界区域编码了一种新的模糊性或愿望形式，作为实现的过程。所有这一切——符号语言的表示能力和逻辑一致性；与人类认知共同进化的技术集合的构建；语言作为人类和计算知识结构之间的桥梁的作用——所有这些都被算法所掩盖，而算法是一种基本上未经检验的结构，我们用它来将想法实例化为计算和文化交叉点的过程。

The development of the Church–Turing thesis and the foundations of effective computability created a new linguistic machinery by means of the Turing machine. This conceptual object, this abstraction engine, opened up a space of operations for computation and served the essential function of clearly articulating the nature of our dependence on limited symbolic systems. Thanks to these proofs, we learned where the boundaries are. The mathematical proofs of effective computability offered by Church, Turing, and others created a new kind of certainty, and a new metaphor for thinking with—the universal mathematical ceiling of computability, and the Turing machine, respectively. But they also encoded a new form of ambiguity, or desire, in the boundary region of effective computability as implemented processes. All of this—the representational power and logical consistency of symbolic language; the construction of technical ensembles that coevolve with human cognition; the role of language as a bridge between human and computational structures of knowledge—all of it gets swept under the rug of the algorithm, the largely unexamined construct we use to instantiate ideas as processes at the intersection of computation and culture.

进程和暂停状态

Process and Halting States

现在我们已经将算法拆解开来，可以在新的语境下将其重新组合。将这些线索串联起来，我们就能看出算法的多重且日益增长的意义，这个看似肤浅且陈旧的文化形象正是本书的核心。算法是一种将符号逻辑结构付诸实践的理念。它植根于计算机科学，但它也充当着一个棱镜，可以反射更广泛的文化、哲学、数学和想象中的语法。67^其中最根本的矛盾存在于算法作为一种有效程序（旨在在可预测的时间内产生答案的一组步骤）与它作为一种永久计算过程的功能之间。实用主义者对算法的定义和计算主义的意识形态中都包含一个关键的假设。在每种情况下，数学解的逻辑抽象都与经过深思熟虑的、具有时间限制的过程定义紧密相连。对于工程师来说，这是一种解决问题的方法。对于丘奇、图灵和计算学家来说，有效可计算问题的概念以及图灵机本身都依赖于处理，即在有限时间内执行指令。在某种程度上，这似乎有些肤浅或重复：方法的定义必须依赖于某种方法的概念。但是，就像图灵机，或者 Chun 对软件的定义一样，方法本身已经成为一种隐喻，或多或少可见，它代表了算法如何作为一个永远运行的过程，持续不断地模拟现实。

Now that we have taken the algorithm apart we can reassemble it with a new context. Drawing these threads together, we can see the multivalent and growing significance of the algorithm, that seemingly shallow and shopworn cultural figure at the heart of this book. The algorithm is an idea that puts structures of symbolic logic into motion. It is rooted in computer science but it serves as a prism for a much broader array of cultural, philosophical, mathematical, and imaginative grammars.⁶⁷ The most radical of these tensions exists between the algorithm’s role as an effective procedure, a set of steps designed to produce an answer in a predictable length of time, on the one hand, and its function as a perpetual computational process, on the other. There is a crucial supposition embedded in both the pragmatist’s definition of the algorithm and the ideology of computationalism. In each case, the logical abstraction of mathematical solutions is yoked to a carefully considered definition of process as time-limited. For engineers it’s a method to solve a problem. For Church, Turing, and the computationalists, the notion of effectively computable problems and the Turing machine itself both depend on processing, on carrying out instructions in finite time. At one level this might seem facile or tautological: the definition of a method must rely on some notion of method. But, like the Turing machine, or Chun’s definition of software, the method has become its own metaphor, more or less visible, for how the algorithm really works as a process that runs forever, persistently modeling reality.

谷歌搜索就是这样一种算法，它将这些矛盾嵌入其技术和文化架构之中。该系统基于多种因素提供相关结果，在百分之一秒内完成执行——这一高效程序自豪地宣称其每次查询都以极快的速度完成。然而，作为一个过程，搜索永无止境地运行，随着谷歌将新的信息源聚合到其系统中，其覆盖范围和影响力也在互联网上不断扩大。正如我们将在下一章中看到的那样，这种影响力也延伸到了未来，因为谷歌除了回答我们当前的问题外，还专注于预测我们未来的需求。对于谷歌而言，可计算问题的空间正在多个维度上不断扩展。谷歌、苹果、亚马逊以及许多其他实体的工程师们正在不懈努力，积极突破有效可计算性的极限，以改进他们的产品，并创造新的文化机器：一个无限前沿，无限处理。

Google search is one such algorithm that embeds these tensions into its technical and cultural architecture. This system delivers relevant results based on a wide range of factors, completing its execution in hundredths of a second—an effective procedure that proudly announces the rapidity of its completion with every query. And yet, as a process, search operates perpetually, extending its reach and influence over the Internet as Google aggregates new sources of information into its systems. As we will see in the next chapter, that influence extends into the future as well, as the company focuses on anticipating our future needs in addition to answering our present questions. For the company, the space of computable questions is continually expanding in multiple dimensions. Engineers at Google, Apple, Amazon, and many other entities are working ceaselessly to actively push the envelope of effective computability in order to make their products better and to create new culture machines: a limitless frontier for limitless processing.

如此一来，算法过程便超越了有效程序的逻辑，成为一种稳态的技术存在，正如西蒙东所言。搜索并非只是一个在瞬间突然启动的系统；它是一个持久而高度复杂的有机体，它同时影响着互联网的形态，推动着机器学习、分布式计算和其他各个领域的创新，并改变着我们自身的认知实践。

In this way the process of the algorithm transcends the logic of the effective procedure to become a steady-state technical being, as Simondon would have it. Search is not just a system that leaps into action for a fraction of a second here or there; it is a persistent, highly complex organism that simultaneously influences the shape of the Internet, drives new innovations in machine learning, distributed computing and various other fields, and modifies our own cognitive practices.

过去十年参与数字文化的任何人都应该清楚，这种现象并非稳态，而是不可逆转地朝着特定目标发展。而且，正如奇点论者所指出的那样，这种发展还在不断加速。从摩尔定律（预测计算机芯片上晶体管的数量每两年翻一番）到全球数据产量的爆炸式增长，“普适计算”显然将继续在物理和文化空间上创造一层越来越厚的传感器、数据和算法。从电视节目到金融，在一个由时间有限（有效）程序和永恒流程之间的张力所驱动的扩张时期，我们正在为计算开辟新的空间。

It should be clear to anyone who has participated in digital culture for the past decade that this phenomenon is not homeostatic, but rather is moving irreversibly toward particular goals. And, as believers in the singularity like to point out, it is accelerating as it goes. From Moore’s Law (predicting that the number of transistors packed onto a computer chip would double every two years) to the explosive growth in global data production, it’s obvious that “ubiquitous computing” will continue to create a thickening layer of sensors, data, and algorithms over physical and cultural space. From television shows to finance, we are claiming new spaces for computation in a period of expansion fueled by the tension between time-limited (effective) procedures and perennial processes.

我们得出的答案是不断扩展问题空间，同时仍然提供有限的计算解决方案。算法思维蕴含着计算主义的愿景，即所有复杂系统最终都将通过计算表征变得等价的极端主义思想。这体现了对有效可计算性的渴望，并对人类的生存产生了深远的影响。随着我们扩展的思维不断构建新的系统、功能、应用程序和操作区域，“人之为人”的问题变得越来越抽象，越来越与代码的隐喻和假设交织在一起。媒体学者安德烈斯·瓦卡里和贝琳达·巴内特在探讨哲学家伯纳德·斯蒂格勒对西蒙东技术观的解读时指出：

The answer we have come up with is to continually expand the problem space while still offering finite computational solutions. Algorithmic thinking encodes the computationalist vision, the maximalist idea that all complex systems will eventually be made equivalent through computational representation. This is the desire for effective computability writ large, and it has existential consequences for humanity. As our extended mind continues to elaborate new systems, functions, applications, and zones of operation, the question of what it means to be human grows increasingly abstract, ever more imbricated in the metaphors and assumptions of code. Discussing Simondon’s vision of technics as interpreted by fellow philosopher Bernard Stiegler, media scholars Andrés Vaccari and Belinda Barnet argue that

两位哲学家都将纯粹人类记忆（以及由此衍生的纯粹思维）的概念置于危机之中，并开启了一种可能，这将激发未来机器人历史学家的兴趣：人类记忆或许只是机械化进程中一个阶段。换句话说，这些未来的机器将以技术存在物的一种补充形式来接近人类记忆（以及由此延伸的文化）^。68

both philosophers put the idea of a pure human memory (and consequently a pure thought) into crisis, and open a possibility which will tickle the interest of future robot historians: the possibility that human memory is a stage in the history of a vast machinic becoming. In other words, these future machines will approach human memory (and by extension culture) as a supplement to technical beings.⁶⁸

我们害怕被思维机器取代的生存焦虑是算法思维的每一个线索的基础，从图灵测试的口号和维纳关于“人之为人”的论证，到数字计算逐渐侵蚀许多人类职业，首先是成为“计算机”。然而，没有什么比在扩展认知的背景下更令人不安的了。随着我们将越来越多的思维外包给算法系统，我们也需要面对依赖于我们无法控制的过程的后果。有令人信服的证据表明，根据社会学家威廉·F·奥格本和多萝西·托马斯的说法，人类记忆和经验的外化使某些技术进步“不可避免”。69^文化本身的通用机器可能会引发新的智力发现，使某些发明不仅成为可能，而且在某些历史关头成为必然。微积分、自然选择、电报：所有这些都以各种形式被多次“发现”或“发明”，如同文字、思想和方法在相关的科学圈子中流传。正如瓦卡里和巴内特关于未来机器人历史学家的戏谑设想所暗示的那样，我们很容易将这些事件解读为漫长进步弧线中的几个瞬间，而人类或许不会走到这一步。

Our existential anxiety about being replaced by our thinking machines underlies every thread of algorithmic thinking, from the shibboleth of the Turing test and Wiener’s argument for the “human use of human beings” to the gradual encroachment of digital computation on many human occupations, beginning with that of being a “computer.” Nowhere is the prospect more unsettling than in the context of extended cognition, however. As we outsource more of our minds to algorithmic systems, we too will need to confront the consequences of dependence on processes beyond our control. There is some compelling evidence to suggest that the externalization of human memory and experience makes certain technological advances “inevitable,” according to sociologists William F. Ogburn and Dorothy Thomas.⁶⁹ The universal machine of culture itself might prime new intellectual discoveries, making certain inventions not just possible but inescapable at certain historical junctures. Calculus, natural selection, the telegraph: all were “discovered” or “invented” multiple times, in various guises, as words, ideas, and methods circulated through the right scientific circles. As Vaccari and Barnet’s playful notion of future robot historians suggests, it’s easy to read these events as moments in a long arc of progress that might not include humanity at its end.

正如斯蒂格勒在回应西蒙东的部分观点时所指出的，能动性的平衡或许已经取决于技术系统：

As Stiegler has argued in partial reply to Simondon, the balance of agency may already lie with technical systems:

如今，机器是工具的承载者，而人类不再是技术个体；人类要么成为机器的仆人，要么成为机器的装配工 [assembliste]：人类与技术对象的关系已被证明发生了深刻的变化^。70

Today, machines are the tool bearers, and the human is no longer a technical individual; the human becomes either the machine’s servant or its assembler [assembliste]: the human’s relation to the technical object proves to have profoundly changed.⁷⁰

这是一个过程的相变。作为运行符号逻辑的载体，算法逐渐不仅仅管理记忆，还管理决策。许多人类领域，尤其是在技术研究领域，日益复杂，加深了我们对计算系统的依赖，并在许多情况下使科学实验本身成为有效可计算性的领域。算法研究方法已经促使一些研究人员认为，“自动化科学”将彻底改变技术进步，甚至可能使假设的生成变得过时，因为算法会不断与海量数据交互。71^算法已经生成了数学证明，甚至是人类无法理解的新的解释方程，使它们“真实”但“不可理解”，数学家史蒂文·斯特罗加茨将这种情况称为“洞察力的终结” ^。72

This is a phase shift in process. As a vessel for putting symbolic logic into motion, the algorithm has increasingly come to manage not just memories but decisions. The growing complexity of many human fields, particularly in technical research, has deepened our dependence on computational systems and in many instances made scientific experimentation itself a domain for effective computability. Algorithmic approaches to research have already prompted some investigators to argue that “automated science” will revolutionize technical progress, perhaps even making the generation of hypotheses obsolete as algorithms continuously interact with huge volumes of data.⁷¹ Algorithms have generated mathematical proofs and even new explanatory equations that defy human comprehension, making them “true” but not “understandable,” a situation that mathematician Steven Strogatz has termed the “end of insight.”⁷²

对斯蒂格勒来说，这是一场噩梦；对另一些人而言，这预示着计算的狂喜，奇点的事件视界，届时算法智能将超越人类（并给我们人类带来臭名昭著的难以预测的后果）。如果说代码的起源始于语言、逻辑以及操纵符号以产生意义，那么这便是其神话般的结局，是符号对意指的胜利。我们称之为算法的巅峰，届时技术变革将加速到足以超越人类智能的速度。在这种情况下，我们不再操纵符号，也无法再解读它们的含义。正如哲学家尼克·博斯特罗姆、计算机科学家弗诺·文格等人所认为的那样，这是计算主义的终局，是对人类与技术关系的一次存在主义公投。⁷³如果我们足够遵循有效程序的渐近线，计算空间的进步不仅会是先锋，而且会是后卫，而人类可能会被抛在后面——不再有效或高效到值得效仿或关注。

For Stiegler this is a nightmare; for others it presages the computational rapture, the event horizon of the singularity, when algorithmic intelligence transcends humanity (with infamously unpredictable results for our species). If the origin story of code begins with language, logos, and the manipulation of symbols to generate meaning, this is its mythical finale, the triumph of sign over signification. We know it as the apotheosis of the algorithm, when technological change will accelerate to such speed that human intelligence may simply be eclipsed. In this scenario we no longer manipulate the symbols and we can no longer construe their meaning. It is the endgame of computationalism as considered by philosopher Nick Bostrom, computer scientist Vernor Vinge, and others, an existential referendum on the relationship between humanity and technics.⁷³ If we follow the asymptote of the effective procedure far enough, the space of computation advances with not just a vanguard but a rearguard, and humanity might simply be left behind—no longer effective or efficient enough to merit emulation or attention.

讽刺的是，这种可能的终极状态——顿悟的终结——是一种理性主义的浪漫，其渊源直指启蒙运动核心中深刻的人文探究精神。它延伸了改变世界的《百科全书》的共同缔造者之一丹尼斯·狄德罗的愿景，正如我们将在第二章中看到的那样：坚持运用启蒙运动的体系或程序最终将导致一种超越知识的状态，而机器人百科全书编纂者能否体验超越则是一个悬而未决的问题。艾萨克·阿西莫夫将这一愿景进一步拓展，在他的《基地》系列故事中称之为“心理史学”。他设想，只要拥有足够的智慧和数据，我们就能预测人类事件的进程，因为文化是算法化的，因为个体和环境可以根据可靠的规则被抽象出来。如果说奇点提供了一种解读计算主义终局的方法，那么这就是第二种：工具理性的胜利，而这种胜利是由我们无法再理解的机器所实现的。

Ironically this possible end-state—the end of insight—is a rationalist romance, drawing its lineage straight to the deeply humanistic spirit of inquiry at the heart of the Enlightenment. It extends the vision of Denis Diderot, one of the cocreators of the world-changing Encyclopédie, as we’ll see in chapter 2: persistently applying the system or procedure of the Enlightenment would eventually lead to a state of transcendent knowledge, leaving open the question of whether robot encyclopedists can experience transcendence. Isaac Asimov took that vision still farther, calling it “psychohistory” in his Foundation stories. With enough cleverness and data, he imagined, we can predict the course of human events because culture is algorithmic, because individuals and circumstances can be abstracted away according to dependable rules. If the singularity provides one way to interpret the endgame of computationalism, this is the second: the triumph of instrumental reason effected by machines we can no longer understand.

我们的技术系统具有特定的政治含义，表达了某些形式的权力，而这些权力往往与计算的解放性修辞相矛盾。大卫·戈伦比亚在《计算的文化逻辑》一书中，对这种政治算计进行了索引，并指出

Our technical systems have specifically political implications, articulating certain forms of power that often contradict the emancipatory rhetoric of computation. David Golumbia indexes this political calculus in The Cultural Logic of Computation, noting how

计算机化往往与相对追求权威、等级森严、且往往政治保守的力量相一致——这些力量为现有的权力形式提供了正当性，而这些权力形式又很容易与工具理性项目相融合^。74

computerization tends to be aligned with relatively authority-seeking, hierarchical, and often politically conservative forces—the forces that justify existing forms of power [in a project that] meshes all too easily with the project of instrumental reason.⁷⁴

阿西莫夫将“心理史学”想象为一种具有解放意义的技术发现，而对哥伦比亚而言，它体现了对“人类和社会经验的很大一部分，或许是全部，都可以通过计算过程来解释”这一政治表述的被动接受。75^其核心是抽象政治，哥伦比亚将其与启蒙运动的工具性理性联系起来。传播学者弗雷德·特纳在 20 世纪 60 年代的学生抗议者身上也发现了同样的焦虑，他们利用打孔卡来对抗行政机器：“我是加州大学的学生。请不要折叠、弯折、旋转或毁坏我。” ⁷⁶这第二个目的也以机器的胜利而告终，但哥伦比亚想象的是一种不同类型的引擎：计算管理和量化所带来的各种国家权力和官僚机构。无论是在奇点宇宙中，还是在哥伦比亚对计算意识形态的解读中，人类的地位充其量都是模棱两可的。

The “psychohistory” that Asimov imagined as a potentially emancipatory technical discovery is, for Golumbia, exemplary of a passive acceptance of the political statement that “a great deal, perhaps all, of human and social experience can be explained via computational processes.”⁷⁵ At its heart, this is about the politics of abstraction, which Golumbia ties to the instrumental reason of the Enlightenment. It is the same anxiety that communications scholar Fred Turner traced in 1960s student protestors who repurposed computer punch cards to battle the administrative machine: “I am a UC student. Please do not fold, bend, spindle or mutilate me.”⁷⁶ This second telos also ends with the triumph of the machine, but what Golumbia imagines is a different sort of engine: the kinds of state power and bureaucracy that computational management and quantification enable. The place of the human is ambiguous at best in both the singularity universe and in Golumbia’s reading of computational ideology.

哥伦比亚在这里代表了一系列批评家的观点，他们认为，至少在我们不干预的情况下，计算文化的实际结果就是强化国家权力。在后面的章节中，我们将更仔细地研究塔尔顿·吉莱斯皮所警告的“虚假个性化”，扩展互联网活动家和作家伊莱·帕里瑟在《过滤泡沫》中的论点，以及媒体理论家亚历山大·加洛韦对协议的政治后果的优雅框架。77^但特纳的《从反主流文化到赛博文化》提供了一个令人信服的观点，即反主流文化的冲动是如何从一开始就融入计算文化的结构中的。黑客形象的起源部分来自于 20 世纪 40 年代工业研究实验室的自由言论；这些设施首先为年轻人创造了不仅可以使用电脑，还可以玩电脑的机会^。78

Golumbia stands in here for a range of critics who argue that the de facto result of computational culture, at least if we do not intervene, is to reinforce state power. In later chapters we will examine more closely the “false personalization” that Tarleton Gillespie cautions against, extending Internet activist and author Eli Pariser’s argument in The Filter Bubble, as well as media theorist Alexander Galloway’s elegant framing of the political consequences of protocol.⁷⁷ But Turner’s From Counterculture to Cyberculture offers a compelling view of how countercultural impulses were woven into the fabric of computational culture from the beginning. The figure of the hacker draws its lineage in part from the freewheeling discourse of the industrial research labs of the 1940s; the facilities that first created the opportunities for young people to not only work but play with computers.⁷⁸

但是，正如加洛韦所说，控制论范式以新的眼光重塑了计算的奇妙魔力。

But, as Galloway has argued, the cybernetic paradigm has recast the playful magic of computation in a new light.

随着非物质劳动意义的日益增强，以及随之而来的对游戏——创造力、创新、新颖性、独特性、灵活性、补充性——的培养和利用，作为一种生产力，游戏将越来越与广泛的社会控制结构紧密联系在一起。今天，我们无疑正在见证游戏作为政治进步，甚至政治中立的终结^。79

With the growing significance of immaterial labor, and the concomitant increase in cultivation and exploitation of play—creativity, innovation, the new, the singular, flexibility, the supplement—as a productive force, play will become more and more linked to broad social structures of control. Today we are no doubt witnessing the end of play as politically progressive, or even politically neutral.⁷⁹

这种转变，即游戏的计算，标志着价值观的根本性巨变，我们将在第四章和第五章中更详细地探讨：过程本身取代了人类的游戏和欢乐体验，成为一种价值。对计算主义的政治批判在此达到了顶峰，其论点是，我们最核心的人类体验——不受约束的想象力和创造力的发挥——正日益落入有效可计算性和计算机制的界限之内。

This shift, the computation of play, signals a fundamental sea change in values that we will address in more detail in chapters 4 and 5: the substitution of process itself as a value for the human experiences of play and joy. The political critique of computationalism reaches its pinnacle here, in the argument that our most central human experiences, the unconstrained play of imagination and creativity, are increasingly falling within the boundaries of effective computability and the regime of computation.

执行

Implementation

当计算的神话开始与物质现实互动时，它便达到了极限。如同UPS的ORION一样，现实世界中的计算既混乱又充满偶然性，需要不断的修改和监督。我将这个问题称为“实现”：即对有效可计算性的渴望如何转化为实际计算机、人类和社会结构的工作系统。了解算法黑匣子内部发生的事情并不能改变一个事实：该操作明确地包含在其实现中：盒子本身同样重要。通过学习解读容器、输入和输出以及实现的接缝，我们可以开始发展一种将算法解读为在代码与文化之间运作的文化机器的方式。

The mythos of computation reaches its limits where it begins to interact with material reality. Like UPS’s ORION, computation in real-world environments is messy and contingent, requiring constant modification and supervision. I call this problem “implementation”: the ways in which the desire for effective computability get translated into working systems of actual computers, humans, and social structures. Learning what goes on inside the black box of the algorithm does not change the fact that the action is specifically contained by its implementation: the box itself is just as important. By learning to interpret the container, the inputs and outputs, the seams of implementation, we can begin to develop a way of reading algorithms as culture machines that operate in the gap between code and culture.

算法正是要弥合这一差距：它们在计算空间和文化空间的交汇处运作，必须在数学和实用的推理模型之间做出妥协或裁定。这项工作的不可避免性，以及算法必须始终被执行才能被使用这一事实，实际上是算法最重要的特征。通过占据并定义这一尴尬的中间地带，算法及其人类合作者扮演了新的角色，成为文化机器，将意识形态与实践、纯粹的数学与不纯粹的人性、逻辑与欲望融为一体。因此，讨论执行，就等于参与一场关于物质性以及那些制定、传输和接收信息的具身主体的对话。80^仅仅提出像 ORION 这样的算法存在于何处，就可能得出非常复杂的答案，涉及由传感器、服务器、员工、代码等等组成的分布式网络。为了解决这个问题，我们需要转向平台研究和媒体考古学，在这些领域，我们可以将执行视为一种物质性形式，其根植于构成“计算表达基础”的硬件和软件^。81

Negotiating that gap is precisely what algorithms do: operating at the intersection of computational and cultural space, they must compromise or adjudicate between mathematical and pragmatic models of reason. The inescapability of that work, the fact that algorithms must always be implemented to be used, is actually their most significant feature. By occupying and defining that awkward middle ground, algorithms and their human collaborators enact new roles as culture machines that unite ideology and practice, pure mathematics and impure humanity, logic and desire. To discuss implementation is thus to join a conversation about materiality and the embodied subjects that enact, transmit, and receive information.⁸⁰ Simply asking the question where an algorithm like ORION exists might lead to very complicated answers involving a distributed network of sensors, servers, employees, code, and so on. To grapple with that question we need to turn to platform studies and media archeology, where we can consider implementation as a form of materiality grounded in the hardware and software that make up the “foundation of computational expression.”⁸¹

在《机制》一书中，媒体学者马修·基尔申鲍姆（Matthew Kirschenbaum）运用两种形式的物质性来解读数字对象。“法医物质性”是指特定数字对象的物理和物质状况：存储数据库的特定硬盘，它拥有自己的机电系统和物理特性。82^在算法领域，数据存储位置和方式的逻辑细节是实现的关键方面，例如维持大型数据中心运行所需的大量能源。“形式物质性”是这些物理计算表现形式所投射的智力阴影。该术语指的是“将多种关系计算状态强加于数据集或数字对象”，或“用户从一组软件逻辑转换到另一组软件逻辑时产生的程序摩擦或感知差异——扭矩”。83^基尔申鲍姆雄辩地描述了观察者或法医调查员在有效解读数字对象方面的重要作用，而形式物质性则说明了有效操作数字对象需要进行多少次转换或调换。

In Mechanisms, media scholar Matthew Kirschenbaum deploys two forms of materiality for reading digital objects. “Forensic materiality” is the physical and material situation of a particular digital object: the particular hard drive where a database is stored, with its own electromechanical systems and physical characteristics.⁸² In the context of algorithms the logistical details of where and how data is stored are vital aspects of implementation, like the vast amount of energy expended to keep major data centers in operation. “Formal materiality” is the intellectual shadow cast by these physical manifestations of computation. The term refers to the “imposition of multiple relational computational states on a data set or digital object,” or the “procedural friction or perceived difference—the torque—as a user shifts from one set of software logics to another.”⁸³ Kirschenbaum eloquently describes the essential role of the observer or forensic investigator in effectively reading digital objects, and formal materiality illustrates how much translation or transposition is involved in effectively manipulating them.

算法系统必须以一种法医化的物质化方式实现，即将其代码和数据存储在某个物理硬盘上，并在某个处理器上运行。但这些物理实例化需要一系列令人眼花缭乱的物质化形式转变，才能被广泛应用：运行该算法的服务器可能实际上是由像 Hadoop 这样的分布式计算平台组织的数百台机器的虚拟集群；算法的各种实例可能存在于由另一个物质化形式层（例如 Docker 平台）管理的软件容器中；算法的公共接口可能会根据用户定制而改变其外观和行为，就像谷歌的定制搜索结果一样。当然，这样的例子不胜枚举：实现的边界似乎无穷无尽，因为它们就是物质宇宙的边界。

An algorithmic system must be implemented in a forensically material way by having its code and data stored on some physical hard drive, running on some processor. But these physical instantiations involve a dizzying flurry of formal materiality moves before they can become broadly accessible: the server running this algorithm might really be a virtual conglomerate of hundreds of machines organized by a distributed computing platform like Hadoop; the various instances of the algorithm might exist in software containers managed by another formal material layer, something like the Docker platform; the public interfaces for the algorithm might vary their appearance and behavior based on user customization, like Google’s tailored search results. Of course, the list can go on: the boundaries of implementation seem endless because they are the boundaries of the material universe.

然而，算法在实施过程中始终受到限制，因为有效可计算性原则是其形式身份的核心。正因如此，我选择使用“实施”一词，而非依赖平台研究或物质性本身的概念。作为一种工具或有效程序，算法是一种通过法证和形式类比、假设和声明性框架编码而存在的工具。理想的实施必须编码或嵌入这些外部性的抽象版本，或将其作为结构化输入来处理，才能成功运行并“解决当前问题”。当然，实施的现实总是会带来新的偶然性、依赖性和复杂性，从而混淆算法作为已实施系统的法证和形式状态之间的界限。与众所周知的黑匣子不同，文化机器实际上是多孔的，在与其他社会技术系统的每个连接点上吸收和挤压文化和计算结构。

And yet the algorithm is always bounded in implementation because the principle of effective computability is central to its formal identity. This is why I choose to use the word “implementation” rather than rely on the concept of platform studies or materiality per se. As a tool or effective procedure, the algorithm is an implement that is coded into existence through a framework of forensic and formal analogies, assumptions, and declarative frameworks. The ideal implementation must encode or embed abstracted versions of those externalities, or deal with them as structured inputs, if it is to operate successfully and “solve the problem” at hand. But of course the reality of implementation always incurs new contingencies, dependencies, and complexity, muddying the ground between the forensic and formal status of the algorithm as an implemented system. Unlike the proverbial black box, the culture machine is actually porous, ingesting and extruding cultural and computational structures at every connection point with other sociotechnical systems.

来自“计算大教堂”的伊恩·博格斯特（Ian Bogost）说：

Ian Bogost, from the “Cathedral of Computation”:

一旦你开始仔细观察，就会发现每个算法都背叛了“单一简单性”和“计算纯粹性”的神话。……一旦你对算法和数据神圣性持怀疑态度，你就不能再将任何计算系统仅仅理解为算法。例如，想想谷歌地图。它不仅仅是通过计算机运行的地图软件——它还涉及地理信息系统、地理定位卫星和转发器、人类驾驶的汽车、车顶全景光学记录系统、国际记录和隐私法、物理和数据网络路由系统，以及网络/移动演示设备。这不是算法文化——它只是一种文化^。84

Once you start looking at them closely, every algorithm betrays the myth of unitary simplicity and computational purity. … Once you adopt skepticism toward the algorithmic- and the data-divine, you can no longer construe any computational system as merely algorithmic. Think about Google Maps, for example. It’s not just mapping software running via computer—it also involves geographical information systems, geolocation satellites and transponders, human-driven automobiles, roof-mounted panoramic optical recording systems, international recording and privacy law, physical- and data-network routing systems, and web/mobile presentational apparatuses. That’s not algorithmic culture—it’s just, well, culture.⁸⁴

然而，戳破计算作为一种特殊宗教体验的幻象，会给我们带来新的问题。它或许“只是文化”，但它正日益被这些平台所改变。放弃代码的魔力并不能改变算法实现对文化体系的普遍影响。相反，它使情况变得更加复杂，抹去了硅谷式计算主义布道的虚假简单性和理想主义。在计算完美的表象之下，我们最终看到的正是博格斯特上文所描述的由相互关联的系统、政策框架、人员、假设、基础设施和界面组成的混乱局面。

Piercing the illusion of computation as an exceptional religious experience, however, leaves us with a new problem. It may be “just, well, culture,” but it is a culture increasingly transformed by these platforms. Giving up the magic of code does not change the pervasive effects of algorithmic implementations on cultural systems. Instead, it complicates the picture, erasing the false simplicity and idealism of Silicon Valley–style computationalist evangelism. What we are left with underneath that facade of computational perfection is exactly the mess of interconnected systems, policy frameworks, people, assumptions, infrastructures, and interfaces that Bogost describes above.

换句话说，实施是双向的——我们构建的每一台文化机器，用于与人类物质性具身世界交互，同时也重构了具身空间，改变了认知和文化实践。更重要的是，之所以如此，是因为实施编码了一种对有效可计算性的渴望的特定表达，而当我们与该系统互动时，我们也会回应这种渴望。算法对普遍知识的追求，反映并满足了我们对自我认知和集体意识的永恒渴望。那些为我们建模、预测和推荐事物的系统的有效性，可以滋养博格斯特所警示的宗教体验，而我们为了感受计算的魔力，愿意接受它们的抽象概念。数字文化的算法模型即使只是部分成功，因为它们对已知宇宙进行了排序，也具有一种诱人的特质。你听着一个几乎完美的流媒体音乐电台，告诉自己，这些歌曲，虽然并非完全正确，但却非常适合此刻，因为一个神奇的算法选择了它们。

In other words, implementation runs both ways—every culture machine we build to interface with the embodied world of human materiality also reconfigures that embodied space, altering cognitive and cultural practices. More important, this happens because implementation encodes a particular formulation of the desire for effective computability, a desire that we reciprocate when we engage with that system. The algorithmic quest for universal knowledge mirrors and feeds our own eternal hunger for self-knowledge and collective awareness. The effectiveness of systems that model, predict, and recommend things to us can feed the religious experience Bogost cautions against, and we willingly accept their abstractions in order to feel the magic of computation. There is a seductive quality to algorithmic models of digital culture even when they are only partially successful because they order the known universe. You listen to a streaming music station that almost gets it right, telling yourself that these songs, not quite the right ones, are perfect for this moment because a magical algorithm selected them.

计算算法或许仅仅被描述为数学，但它们却如同文化机器般运作，极大地改变了人类反思的版图，正如我们将在本章后续的算法解读中看到的那样。它们重塑了我们看待自身的空间。我们在现实和虚拟的信息与交换系统中留下的字面和隐喻的足迹，被用于通过定制搜索结果、推荐和其他自适应系统（即帕里瑟所说的“过滤泡沫”）塑造未来的视野^。85

Computational algorithms may be presented as merely mathematical, but they are operating as culture machines that dramatically revise the geography of human reflexivity, as we will see in the algorithmic readings that follow this chapter. They reshape the spaces within which we see ourselves. Our literal and metaphorical footprints through real and virtual systems of information and exchange are used to shape the horizon ahead through tailored search results, recommendations, and other adaptive systems, or what Pariser calls the “filter bubble.”⁸⁵

但当算法从预测跨越到决策，从建模跨越到构建文化结构时，我们发现自己正在修改现实以适应它们之间的差异。任何依赖于抽象的系统中，都会存在剩余，一组被丢弃的信息——延异，或者说，地图与领土之间至关重要的意义区别和延展。这种差距在实施过程中显现，此时必须解决算法的计算理解和文化理解之间的冲突。在许多方面，这种差距为算法的形象创造了文化空间，提供了我们想象中的代码魔法——故障、难以解释的结果和奇异的意外发现。在《雪崩》中，实施问题支撑了几个关键的情节转折，但最令人难忘的莫过于在多元宇宙中绘制三维空间，多元宇宙是主角 Hiro、其他黑客和其他 Technorati 成员聚集的虚拟现实。Hiro 可以通过将武士刀刺穿墙壁并跟随他的剑来穿墙，利用

But when algorithms cross the threshold from prediction to determination, from modeling to building cultural structures, we find ourselves revising reality to accommodate their discrepancies. In any system dependent on abstraction there is a remainder, a set of discarded information—the différance, or the crucial distinction and deferral of meaning that goes on between the map and the territory. This gap emerges in implementation, when the collisions between computational and cultural understandings of algorithms must be resolved. In many ways the gap creates the cultural space for the figure of the algorithm, providing the glitches, inexplicable results, and strange serendipity we imagine as the magic of code. In Snow Crash, the problem of implementation underwrites several crucial plot twists, but one of the most memorable is the plotting of three-dimensional space in the Multiverse, the virtual reality where Hiro Protagonist, fellow hackers, and other Technorati congregate. Hiro can move through walls by sticking his katana through them and following his sword, exploiting a

几年前，他试图将剑术规则移植到现有的元宇宙软件上时发现了这个漏洞。……但就像元宇宙中的其他任何东西一样，（控制墙壁运作的规则）只不过是一种协议，一种不同计算机同意遵循的惯例。理论上，它不容忽视。但在实践中，它取决于不同计算机在恰当的时间以极高的速度精确交换信息的能力^。86

loophole that he found years ago when he was trying to graft the sword-fighting rules onto the existing Metaverse software. … But like anything else in the Metaverse, [the rule governing how walls function] is nothing but a protocol, a convention that different computers agree to follow. In theory, it cannot be ignored. But in practice, it depends on the ability of different computers to swap information very precisely, at high speed, and at just the right times.⁸⁶

对小说而言，这只是一个方便的伎俩，就像黑客利用或创造的许多东西一样：一种规避标准控制结构的机制，让人想起《议定书》中加洛韦的号召。但就我们的目的而言，它也展现了计算隐喻与文化隐喻、抽象与实现之间差距的本质特征。

For the novel, this is a convenient trick, like many of the things hackers exploit or create: a mechanism for sidestepping standard structures of control reminiscent of Galloway’s call to arms in Protocol. But for our purposes, it also illustrates the essential features of the gap between computational and cultural metaphors, between abstraction and implementation.

Hiro 的武士刀刺法之所以有效，部分原因在于它利用了不同逻辑抽象体系之间的鸿沟——元宇宙中控制剑的算法规则，以及控制虚拟角色和结构的类似规则。此外，它还依赖于计算系统中时间性的抽象建构——正如 Stephenson 指出的那样，这种差距既是空间上的，也是时间上的，取决于 Hiro 的卫星连接与元宇宙中处理其会话的服务器之间的延迟。当 Hiro 利用两个算法系统之间的延迟，真正侵入黑匣子时，他实际上进行了一种套利。最后，这种差距是文化上的：当 Hiro 将剑刺入墙壁时，他提出了一个不可能的问题，而他得到的正是他所希望的不可能答案——“沙赞”，黑客魔法上演了。

Hiro’s katana thrust works in part because it exploits the gulf between different logical regimes of abstraction—the algorithmic rules governing swords in the Metaverse and the set of similar rules governing avatars and structures. It also depends on the abstracted construction of temporality in computational systems—as Stephenson points out, the gap is temporal as much as it spatial, depending on the lag between Hiro’s satellite connection and the servers handling his session in the Metaverse. Hiro engages in a kind of arbitrage when he exploits the lag between two algorithmic systems to literally hack his way into a black box. And, finally, the gap is cultural: Hiro asks an impossible question when he pokes his sword into the wall, and he receives just the impossible answer he was hoping for—shazam, hacker magic is performed.

然而，重要的是要认识到，这种差距不同于故障、崩溃或其他计算系统故障的迹象。计算全知的表象崩塌的时刻，有助于我们看清差距，并催生了引人入胜的计算机艺术和表演流派，但它们仅仅是通往计算与现实之间更广阔空间的窗口。我们在计算和现实的两端构建差距，或为其创造空间。算法系统和计算模型用各种抽象的姿态忽略了复杂系统的关键方面，而它们留下的东西则不安地徘徊在边缘地带，既已知又未知，既被理解又被遗忘。但是，当我们围绕计算系统组织新的认知模式，并选择忘记或放弃我们曾经拥有的知识形式时，这些系统的人类参与者、用户和架构师在构建差距方面也扮演着同样重要的角色。每一个依赖的时刻，比如我们现在依赖算法系统提供的忘记的电话号码或拼写，特别是每一个被拒绝的直接、非中介的人际接触的机会，都会在计算和人类经验之间增加一点空间。

It is important to realize, however, that the gap is not the same as the glitch, the crash, or other signs of malfunctioning computational systems. These moments when the facade of computational omniscience falls are very helpful in seeing the gap, and they have given rise to fascinating genres of computer art and performance, but they are only windows into the broader opening between computation and reality. We construct the gap, or create space for it, on both sides. Algorithmic systems and computational models elide away crucial aspects of complex systems with various abstracting gestures, and the things they leave behind reside uneasily in limbo, known and unknown, understood and forgotten at the same time. But the human participants, users, and architects of these systems play an equally important role in constructing the gap when we organize new cognitive patterns around computational systems and choose to forget or abandon forms of knowledge we once possessed. Every moment of dependence, like a forgotten phone number or spelling that we now depend on an algorithmic system to supply, and especially every rejected opportunity for direct, unmediated human contact, adds a little to the space between computation and human experience.

算法是抽象、咒语、数学和技术记忆的复杂集合。它们是计算大教堂的物质实现。87^当我们与它们互动时，我们是在与神谕、神明和小恶魔对话，讨论着一种充满命令词、布尔连接词以及常常包含深度个人信息的洋泾浜语或行业语言。通过这些互动，我们不断地重塑算法的神话，通过背诵熟悉的咒语（喃喃自语“OK Google Now”或输入熟悉的网址）来重申它，并随着我们与计算文化机器建立更复杂的关系而扩大其影响范围。这些关系依赖于多种形式的素养——我们现在都在解读算法系统，或多或少，这取决于我们对实现环境的认识和关注。

Algorithms work as complex aggregates of abstraction, incantation, mathematics, and technical memory. They are material implementations of the cathedral of computation.⁸⁷ When we interact with them, we are speaking to oracles, gods, and minor demons, hashing out a pidgin or trade language filled with command words, Boolean conjunctions, and quite often, deeply personal information. We are constantly reworking the myth of the algorithm through these interactions, reaffirming it through our recitals of the familiar invocations (muttering “OK Google Now” or tapping out a familiar URL) and extending its reach as we develop more sophisticated relationships with computational culture machines. Those relationships depend on multiple forms of literacy—we are all reading algorithmic systems now, more or less badly, depending on our awareness and attention to the context of implementation.

算法阅读

Algorithmic Reading

为了有效地解读算法那奇特的形象，即刻意刻意营造的毫无特色、如同变色龙般穿梭于计算逻辑与文化逻辑之间的通道，我们需要自己采取一种算法的方法。解读复杂的计算文化对象需要一套有效的程序，该程序在批判理论、计算逻辑和文化理解之间的实施空间中运作。正如计算算法蕴含着使万物有效可计算的愿望一样，我们也应该认识到算法阅读带来的议程：使计算的各个方面都为人类所理解的愿望。正如文学评论家斯蒂芬·拉姆齐在《阅读机器》的结论中所说，他称之为算法批评的“新型批评行为”不仅是可能的，而且是必要的，“隐含在众多旨在促进思考、自我表达和社群的界面中”。88^算法阅读，正如我在下文中定义的，是一种思维模式或思考工具，任何人都可以使用它来解读文化产物。

To effectively read the strange figure of the algorithm, that deliberately featureless, chameleon-like passthrough between computational and cultural logics, we need to take an algorithmic approach ourselves. The reading of complex computational cultural objects requires its own effective procedure, one that operates in the space of implementation between critical theory, computational logic, and cultural understanding. Just as computational algorithms embed a desire to make all things effectively computable, we should recognize the agenda that algorithmic reading brings with it: a desire to make all facets of computation legible to human beings. As literary critic Stephen Ramsey argues in the conclusion to Reading Machines, the “new kinds of critical acts” he terms algorithmic criticism may be not only possible but necessary, “implicit in the many interfaces that seek only to facilitate thought, self expression [sic] and community.”⁸⁸ Algorithmic reading, as I define it below, is a mode of thought, or a tool for thinking, that anyone can use to interpret cultural artifacts.

从这个角度来看，算法阅读在相互竞争的欲望之间形成三角关系：一方面是计算主义者不断拓展有效程序边界的追求，另一方面是人类对普遍知识的渴望。两者之间，存在着一种我们才刚刚开始认识到的新事物：一种相互构成的欲望，即创造和操纵差距，让某种魔力从抽象与实现的复杂互动中涌现，就像计算生命游戏中的鸟群一样。这种延异为我们与计算之间不断发展的热爱提供了能量，也是我们在进行算法阅读时所利用的资源。

In this light algorithmic reading triangulates between competing desires: the computationalist quest to continually expand the boundary of the effective procedure, on the one hand, and the human desire for universal knowledge, on the other. Between them, something new that we are only now beginning to recognize: the mutually constitutive desire to create and manipulate the gap, to have a kind of magic emerge from the complex interactions of abstraction and implementation like flocks of birds from a computational game of life. That différance provides the energy for our evolving love affair with computation, and it is the resource we tap into when we perform algorithmic readings.

这就是为什么解读差距并重建构成每个文化机器、每个多孔计算盒子的墙壁和联系的计算和社会力量本身就是一种“解决问题的方法”。与算法本身一样，算法解读是一个复杂的概念结构，包含多层流程、抽象和与现实的接口。

This is why reading the gap and reconstructing the computational and social forces that make up the walls and linkages of each culture machine, each porous computational box, is itself a “method for solving a problem.” Like the algorithm itself, algorithmic reading is a complex conceptual structure containing layers of processes, abstractions, and interfaces with reality.

算法研究的对象远远超出了特定文本片段或多媒体的表面表现。阅读Facebook上的某条帖子，甚至阅读文学学者Jeff Nunokawa在Facebook上发表的散文集《Note Book》，除非直接与Facebook本身的机制互动，否则只能捕捉到合作中人性化的一面。因此，算法阅读借鉴了我们在此已经探讨过的多个重要先驱——控制论、文化研究、平台与软件研究、媒体理论以及数字物质性。我们才刚刚开始研究如何将这些不同的视角整合在一起，以探讨算法的伦理、软件的可读性和计算的政治性。算法平台如今有效地塑造了所有文化产品，从作者为了推广新书而进行的例行推特玩笑，到向我们推荐新产品的复杂系统。算法阅读的核心原则，也是该方法的与众不同之处，在于我们必须将文化机器本身作为研究对象，而不仅仅是其文化产出。为了有效地做到这一点，我想提供一组关键术语或变革概念，它们是任何文化机器的核心功能。

The algorithmic object of study extends far beyond the surface manifestation of a particular fragment of text or multimedia. A reading of a particular post on Facebook, or even, say, Note Book, a collection of literary scholar Jeff Nunokawa’s essayistic Facebook posts, would capture only the human side of the collaboration unless it engaged directly with the apparatus of Facebook itself. In this way algorithmic reading draws from the multiple critical forerunners we have already considered here—cybernetics, cultural studies, platform and software studies, media theory, and digital materiality. We are just beginning to work out how to pull these different perspectives together to ask questions about the ethics of algorithms, the legibility of software and the politics of computation. Algorithmic platforms now shape effectively all cultural production, from authors engaging in obligatory Twitter badinage to promote their new books to the sophisticated systems recommending new products to us. A central tenet of algorithmic reading, what distinguishes the method, is that we must take the culture machine itself as the object of study, rather than just its cultural outputs. To do that effectively, I’d like to offer a set of key terms or transformative concepts that serve as the central functions of any culture machine.

我们需要的第一个方法论工具是对过程有扎实的批判性理解。各种算法都推进了有效可计算性论证的某种版本，它们或明或暗地编码了论证，表明问题——无论是农业问题还是平方根的开方——都可以通过遵循该方法的步骤来解决。这样，过程本身就是一种批判性理解的排序逻辑，它依赖于马丁·海德格尔、阿尔弗雷德·怀特海、西蒙东和斯蒂格勒等哲学家所倡导的“过程哲学”概念。算法的研究对象是一个运动中的系统，是随着时间推移而形成的一系列迭代。算法系统最重要的方面不是它在任何特定时刻呈现给世界的表面材料（例如，出现在某人 Facebook 动态顶部的内容），而是不断生成和操纵这些表面材料的规则和代理系统（例如，过滤和推广特定内容的算法）。正如我们上面所探讨的，这个过程嵌入了自我延续与完成之间的张力，以及优雅结束的有效程序与充满宇宙的通用计算精神之间的张力。

The first methodological tool we need is a grounded critical understanding of process. Algorithms of all kinds advance a version of the effective computability argument, encoding explicit or implicit arguments that the problem—whether it is agriculture or square root extraction—can be solved by following the steps of the method. In this way process itself is an ordering logic for critical understanding, leaning on notions of “process philosophy” espoused by the philosophers Martin Heidegger, Alfred Whitehead, Simondon, and Stiegler, among others. The algorithmic object of study is a system in motion, a sequence of iterations that comes into being as it moves through time. The most important aspect of an algorithmic system is not the surface material it presents to the world at any particular moment (e.g., the items appearing at the top of one’s Facebook feed) but rather the system of rules and agents that constantly generate and manipulate that surface material (e.g., the algorithms filtering and promoting particular nuggets of content). That process embeds, as we explored above, the tension between self-perpetuation and completion, between an effective procedure that ends gracefully and a spirit of universal computation that fills the universe.

这种过程概念与我们的第二个方法论关键词“抽象”密切相关。算法系统之所以成为研究对象，不仅在于其包含的内容，还在于其省略的内容。抽象系统将电信号转化为汇编语言、高级代码、图形用户界面，再转化为图标和文化隐喻系统（中间还包含许多其他层次），从而构建了关于现实的意识形态框架和论点。像海尔斯、加洛韦和麦肯齐·沃克这样的媒体学者的研究阐明了这些抽象概念如何在现实世界中运作。如果说算法是文化机器，那么抽象就是它们的主要输出之一。例如，在第四章中，我讨论了优步的应用程序界面，其中一张卡通地图显示了汽车在城市网格中穿梭。优步依赖于抽象化打车的复杂性、规章制度和既定惯例，将打车体验变成了一种电子游戏。这种抽象模式非常成功，以至于硅谷的一整类初创企业现在可以被归类为“X 版优步”，而这个 X 实际上是一种双重抽象。首先，我们将Uber简化的、自由代理的“共享经济”商业模式应用于另一个经济领域。然后，我们让所有这些领域都具有可互换性，一个变量X可以代表市场上任何一个角落，只要这些角落的普适计算和算法服务尚未颠覆现状。如同图灵最初的抽象机一样，这些系统将一种符号逻辑扩展到文化世界，重新排序与之接触的思想和意义。

This notion of process depends intimately on our second methodological keyword, abstraction. Algorithmic systems are objects of study not only for what they include, but for what is elided. The systems of abstraction that translate electrical signals to assembly language to high-level code to a graphical user interface to a system of icons and cultural metaphors (with many other layers in between) create ideological frames and arguments about reality. The work of media scholars like Hayles, Galloway, and McKenzie Wark serve to illuminate how these abstractions work in the world. If algorithms are culture machines, abstractions are one of their primary outputs. As an example, in chapter 4 I discuss Uber’s application interface, with a cartoonish map showing cars roaming the urban grid. Uber depends on abstracting away the complexities, regulations, and established conventions of hailing a cab, turning the hired car experience into a kind of videogame. That mode of abstraction has been so successful that an entire genre of Silicon Valley startups can now be categorized as “Uber for X,” where that X is actually a double abstraction. First, we adapt Uber’s simplifying, free-agent “sharing economy” business model to another economic arena. Then we make all such arenas fungible, a variable X that can stand for any corner of the marketplace where ubiquitous computing and algorithmic services have yet to disrupt the status quo. Like Turing’s original abstraction machine, these systems extend a symbolic logic into the cultural universe that reorders minds and meanings that come into contact with them.

这些交互的媒介是我们的第三个关键词，即实现状态。文化机器运行的流程及其产生的抽象只能存在于实现空间中。这个空间是计算和文化对现实的建构之间的差距，文化机器既生成又操纵这个差距，以实现其程序目标和有效可计算性的更广泛扩展。Netflix 决定使用一组人类“标记者”根据一系列定性和定量指标来评估其流媒体视频目录，这代表了实现上的深刻转变，正如我们将在第三章中看到的那样。他们抛弃了第一个推荐算法的纯统计方法，转而采用更混乱、更具文化纠缠的过程，这种转变现在影响着基于人工和算法输入创作像《纸牌屋》这样的原创内容的更为复杂的业务。

The medium for these interactions is our third keyword, the state of implementation. The processes that culture machines run and the abstractions they produce can only exist in the space of implementation. That space is a gap between computational and cultural constructions of reality, one that culture machines both generate and manipulate in order to achieve their procedural objectives and the broader expansion of effective computability. Netflix’s decision to use a group of human “taggers” to evaluate its streaming video catalog according to a range of qualitative and quantitative metrics represented a profound shift in implementation, as we’ll see in chapter 3. They left behind the purely statistical approach of their first recommendation algorithm in favor of a messier, more culturally entangled process, a transformation that now informs the even more complicated business of creating original content like House of Cards based on human and algorithmic inputs.

在创造具有文化复杂性和审美评价的文化机器和创意作品方面，人类与算法之间的相互依赖性日益加深，这让我们看到了算法阅读中最具挑战性的方面：想象力。计算与文化之间的差距不仅仅是不同符号逻辑、表征和意义系统之间的鸿沟，也是不同想象模式之间的差距。所有符号系统、所有语言都包含一种特定的可能性逻辑，一种取决于表征性质和语义关系的想象范围。数学家可以精确描述高度抽象的关系，而这些关系几乎不可能用更熟悉的人类语言来定义。计算系统正在开发新的想象思维能力，这可能从根本上与人类的认知格格不入，包括从数百万个统计变量中创建推论，以及在随机、快速变化的情况下操纵系统，而这些情况在时间上落后于我们有效理解的能力。在接下来的阅读中，我们可以看到计算想象力的身影，从 Netflix 副总裁在他自己的系统结果中描述的“机器中的幽灵”，到我们都曾在计算的完美功能和可预测性外表的边缘瞥见过的各种奇怪的意外发现和美丽的故障^。89

The growing interdependence of humans and algorithms in creating culturally complex, aesthetically evaluated culture machines and creative works leads us to the most challenging aspect of algorithmic reading: imagination. The gap between computation and culture is not just a gulf between different systems of symbolic logic, of representation and meaning: it is also a gap between different modes of imagination. All symbolic systems, all languages, contain a particular logic of possibility, a horizon of imagination that depends on the nature of representation and semantic relationships. Mathematicians can precisely describe highly abstract relationships that are almost impossible to define in more familiar human language. Computational systems are developing new capacities for imaginative thinking that may be fundamentally alien to human cognition, including the creation of inferences from millions of statistical variables and the manipulation of systems in stochastic, rapidly changing circumstances that are temporally behind our ability to effectively comprehend. We see computational imagination throughout the readings that follow, from the “ghost in the machine” that a Netflix VP described in his own system’s results to the kinds of strange serendipity and beautiful glitches we have all glimpsed at the edges of computation’s facade of perfect functionality and predictability.⁸⁹

综上所述，算法阅读的这些组成部分构成了一种新的配方，一种以算法的方式理解算法的文化方法。它是一种借助机器的光影进行阅读的方式：计算增强认知的璀璨光芒和黑匣子的迷惑。正如我们的关键词所示，算法阅读是一个关键框架，它帮助我们解读那些也在解读我们的对象：计算系统在相互的诠释过程中适应我们的行为。

Taken together, these components of algorithmic reading provide the ingredients for a new recipe, an algorithmic approach to the cultural understanding of algorithms. It is a means of reading by the lights and shadows of machines: the brilliant illumination of computationally enhanced cognition and the obfuscations of black boxes. As our keywords suggest, algorithmic reading is a critical frame for interpreting objects that are also interpreting you: computational systems that adapt to your behavior in a mutual hermeneutic process.

毕竟，我们已经在与算法交流——分享、信任并委托它们代表我们思考和行动。在这次挖掘中，我挖掘出的每一个邪恶的黑匣子和压迫性平台，都蕴藏着闪光点：令人惊叹的创造力和洞察力，如果没有人机的协作，这些都不可能实现。如同我们所有其他的神话一样，文化机器始终是我们自己。我们制造这些工具，赋予它们力量和历史，因为我们试图在脆弱的人类躯体之外，保卫自身的某些部分。我们建造大教堂、仪式和集体故事，为自己施展魔法，建造一座永恒记忆的圣殿，以留住我们最闪耀的时刻。理解算法的形象是成为真正合作者的第一步——不仅与机器合作，也通过算法系统所构建的庞大集体，与彼此合作。在所有这些层层堆积的符号、代码和逻辑之下，我们发现算法的形态并非静止不动，而是动态的。算法解读需要在计算与文化之间充满活力的互动领域中运作。这是算法想象力的游乐场，是人类与计算组合能够创造非凡而美妙事物的天地。

After all, we are already communing with algorithms—sharing, trusting, and deputizing them to think and act on our behalf. For every nefarious black box and oppressive platform I unearth in this dig, there are bright spots: instances of astounding creativity and insight that would never have been possible without the collaboration of human and machine. Like all of our other myths, the culture machine has been us all along. We build these tools, we imbue them with power and history, because we seek to secure some part of ourselves outside the fragile vessel of the human form. We build cathedrals, rituals, and collective stories to cast a spell on ourselves, a nam-shub of eternal memory to keep our brightest moments alive. Understanding the figure of the algorithm is the first step to becoming true collaborators—and not just with machines, but with one another through the vast collectives that algorithmic systems make possible. Underneath all these layers of silted symbol, code, and logic, we find that the figure of the algorithm is not fixed but in motion, and that algorithmic reading requires working in a charged sphere of action between computation and culture. This is the playground of algorithmic imagination, the zone where human and computational assemblages can do extraordinary, beautiful things.

笔记

Notes

算法副官

The Algorithmic Aide-de-camp

2011 年 10 月，苹果公司宣布推出 Siri，这是一款“只需发出指令就能帮你完成任务的智能助手” ^。3这款全球领先智能手机的新功能甚至没有成为新闻稿的头条新闻，但它对我们与数字系统的交互产生了深远的影响。

In October 2011, Apple announced the launch of Siri, an “intelligent assistant that helps you get things done just by asking.”³ This new functionality for the world’s leading smartphone did not even make the headline of the press release, but it has had profound implications for our interactions with digital systems.

苹果发布 Siri 似乎预示着一项全新技术的到来，这也是该公司仅有的几项以“测试版”形式发布的技术之一。但 Siri 及其前身已通过美国政府有史以来资助的最大人工智能项目开发了近十年，从国防高级研究计划局 (DARPA) 获得了超过 1.5 亿美元的资助。4^该项目由斯坦福国际研究院 (SRI International，原斯坦福研究所) 负责，汇集了来自几所主要大学和企业的数百名研究人员，共同研究人工智能长期以来的圣杯：对话计算机。5 ^SRI是诺伯特·维纳和控制论专家在二战期间帮助开创的大型跨学科政府研究实验室浪潮的一部分。Siri 的前身被称为“会学习和组织的认知代理”，简称 CALO，旨在帮助战地指挥官管理复杂的数据和官僚机构，同时专注于战略挑战。 CALO 的灵感来自拉丁语calonis，即士兵的仆人，这为其想象中的文化角色赋予了非常不同的色彩：更像是执行官而不是私人助理^。6

Apple’s unveiling of Siri seemed to herald a brand new technology, one of the only things the company has ever launched as a “beta” release. But Siri and its antecedents had been in development for nearly a decade through the largest artificial intelligence project ever funded by the U.S. government, drawing over $150 million from the Defense Advanced Research Projects Agency (DARPA).⁴ Housed at SRI International (originally the Stanford Research Institute), the project brought together hundreds of researchers from several major universities and corporations to work on what has long been a holy grail for artificial intelligence: a conversational computer.⁵ SRI was part of the same wave of large-scale, interdisciplinary government research laboratories that Norbert Wiener and the cyberneticians helped to pioneer during World War II. Dubbed the Cognitive Agent that Learns and Organizes, or CALO, Siri’s precursor was designed to help field commanders who have to manage complex data and bureaucracy while remaining focused on strategic challenges. CALO was inspired by the Latin calonis, or soldier’s servant, giving a very different cast to its imagined cultural role: more executive officer than personal assistant.⁶

CALO 的开发团队克服了诸多技术挑战，例如开发一款能够更好地组织会议的工具。正如科技记者 Bianca Bosker 在她关于 Siri 发展历程的简短文章中写道：

The teams that worked on CALO approached a number of different technical challenges, such as developing a tool to better organize meetings. As technology journalist Bianca Bosker writes in her short history of Siri’s evolution:

假设你的同事在会议前不久取消了会议。CALO 了解每个人在项目中的角色，可以判断是否取消会议，并在必要时重新安排会议、发出新的邀请并确定会议室。如果会议按计划进行，CALO 可以收集（并排序）所有你需要的文件和电子邮件，以便你快速了解当前主题。助理会旁听会议内容，并在会后提供一份打印的记录，记录与会者的发言内容，并概述谈话中安排的具体任务。CALO 还可以帮助制作演示文稿、将文件整理到文件夹中、对收到的消息进行排序、自动生成费用报告，以及执行其他一系列任务^。7

Say your colleague canceled shortly before a meeting. CALO, knowledgeable about each person’s role on a project, could discern whether to cancel the meeting, and if needed, reschedule, issue new invitations and pin down a conference room. If the meeting went ahead as planned, CALO could assemble (and rank) all the documents and emails you’d need to be up to speed on the topic at hand. The assistant would listen in on the meeting, and, afterward, deliver a typed transcript of who said what and outline any specific tasks laid out during the conversation. CALO was also able to help put together presentations, organize files into folders, sort incoming messages and automate expense reports, among a host of other tasks.⁷

这些功能以及其他功能展现了 Siri 的愿景：它将成为一种新型的智能助手，不仅能够自动执行任务，还能有效地预测用户的需求并采取行动。DARPA 的长期投资实现了早期人工智能研究的伟大抱负之一：打造一个能够与人类协同行动的智能代理。

These and other features presented a vision for Siri as a new kind of intelligent aide-de-camp that would not merely automate tasks but effectively predict and take action on the user’s behalf. The long-term investment from DARPA realized one of the great ambitions of an earlier generation of AI research: an intelligent agent that can act for itself in concert with humans.

CALO/Siri 是当代文化机器的一个有力例证：智能代理被设想为一组相互关联的计算系统，是一项雄心勃勃的尝试，旨在使广阔的全新文化领域“有效地可计算”。它体现了一种实用主义的理想，能够克服多项技术挑战：对话式计算机必须能够理解语音查询并做出恰当的回答。它必须能够有效地解析电子邮件、任务列表、预算、日历，以及随着 Siri 功能的扩展，餐厅列表、电影放映时间以及其他众多数据源。作为副官或虚拟助理，Siri 致力于将记忆外化，提醒我们即将举行的会议，并处理其他低级认知任务。在所有这些功能中，Siri 都遵循着相当有限的可能问题和答案脚本，将不符合其宇宙模型的查询转储到搜索框中，并提供来自网络的结果。

CALO/Siri is a powerful example of the contemporary culture machine: conceived as an interlocking set of computational systems, the intelligent agent is an ambitious effort to make vast new swaths of culture “effectively computable.” It embodies a pragmatist ideal for conquering several technical challenges: a conversational computer must interpret spoken queries and answer appropriately. It must effectively parse emails, task lists, budgets, calendars, and, as Siri expanded, restaurant listings, movie times, and many other data sources. As an aide-de-camp or virtual assistant, Siri is a literal effort to externalize memory, reminding us of upcoming meetings and handling other low-level cognitive tasks. In all of these functions, Siri follows a fairly limited script of possible questions and answers, dumping queries that do not fit its model of the universe into a search box and providing results from the Web.

解读 Siri 需要揭开其计算表象的帷幕，但这也意味着解读其在 SRI 的最初设计者以及在苹果公司实施该系统的工程师和营销人员的意图之外的内涵。从这个意义上讲，它是一个整体，一个新兴的“技术存在”，诠释了吉尔伯特·西蒙东（Gilbert Simondon）的技术性概念。换句话说，Siri 是一台文化机器，它不仅塑造我们的互动，整合（并依赖于我们）功能，还肩负着构建和组织宇宙的更广阔的野心。正如西蒙东所说：“技术现实在具备调节性之后，将能够融入文化，而文化本质上也具有调节性。” ⁸ Siri 是处于这一演变过程中间的一种算法，它的技术功能和文化前提让我们能够清晰地看到计算与文化之间的差距，尤其是在我们如何定义知识​​方面。

Reading Siri requires stepping behind the curtain of its computational facade, but it also involves reading beyond the intentions of its original designers at SRI and the engineers and marketers who implemented the system at Apple. In this way it is an ensemble, an emergent “technical being” that illustrates Gilbert Simondon’s notions of technicity. Put another way, Siri is a culture machine that not only shapes our interactions, integrating (and depending on us) for functionality, but also has a broader ambition to structure and organize the universe. As Simondon puts it: “Technical reality, having become regulative, will be able to be incorporated into culture, which is essentially regulative.”⁸ Siri is an algorithm midway through this process of becoming, where its technical functions and cultural premise give us a good view of the gap between computation and culture, particularly in terms of how we define knowledge.

执行知识

Performing Knowledge

许多苹果用户并不知道，在苹果收购其算法之前，Siri 要智能得多。2010 年，这家最初开发 Siri 商业产品的初创公司将 Siri 发布到 iOS 应用商店，几周后，史蒂夫·乔布斯就促成了 Siri 的收购。乔布斯迅速采取行动，抢先与 Verizon 达成协议，将 Siri 软件引入安卓手机。Siri 团队也开始调整其工具，以适应全球数百万用户的需求，这些用户很快就会将其用作 iOS 操作系统的核心界面。9^最初的算法旨在灵活地与数百个数据源交互，综合来自多个档案库的信息来回答你关于附近最好的披萨店的问题，或者在你的航班延误时建议其他交通方式。Siri 的部分功能已逐渐恢复，但与苹果的合作也降低了它的灵活性，因为任何关于该工具提取第三方数据的新协议都需要经过律师审核和签订繁琐的合同。初代 Siri 的对话也更加尖锐，偶尔会说出脏话，而且态度鲜明，这也是它吸引乔布斯的部分原因（尽管苹果公司内部很快就降低了该软件的影响力）。10^该软件不只是知道事情，它还知道一切。

Many Apple users do not know that Siri was a lot smarter before Apple purchased the algorithm. Just a few weeks after the startup that first developed Siri as a commercial product released it to the iOS App Store in 2010, Steve Jobs orchestrated the company’s acquisition. He moved quickly, preempting a deal with Verizon to bring the software to Android phones instead, and the Siri team began the process of adapting their tool for the millions of users across the world who would soon be using it as a core interface for the iOS operating system.⁹ The original algorithm was designed to flexibly interact with hundreds of data sources, synthesizing information from multiple archives to answer your question about the best pizza nearby, or suggesting alternate modes of transportation if your flight was delayed. Some of this functionality has gradually been restored to Siri, but its association with Apple has also made it less flexible, as lawyers and elaborate contracts precede any new agreements for the tool to pull in third-party data. The original Siri also had a sharper edge to its dialog, occasionally deploying four-letter words and plenty of attitude, which was part of its appeal for Jobs (though Apple quickly dialed down the software’s affect in-house).¹⁰ The software didn’t just know things, it was also knowing.

Siri 从一开始就被如此营销，被誉为一台精密得近乎魔法的文化机器：一台不仅能说话，还能理解的电脑。广告中，塞缪尔·杰克逊、佐伊·丹斯切尔和马丁·斯科塞斯与他们的手机进行真实的对话。苹果向我们展示了我们可以扮演人类角色，开着玩笑，用我们正常而独特的方式提问，而 Siri 也能跟上我们的节奏，创造自己的笑话，不仅理解我们的话语，更能理解其背后的含义。广告暗示了程序员们通过他们在 Siri 中留下的“彩蛋”来调情的那种关系，比如那个众所周知的技巧：询问 Siri 把尸体扔到哪里（Siri 给出了一个冷面幽默的概述，比如当地的沼泽地、金属铸造厂等等——尽管苹果后来改变了答案）。这些预先编程的火花照亮了远期目标，即不仅创造智能，更创造真正的个性。然而，这些技巧都有一个我们必须学习的脚本——为了让 Siri 说出每一个妙语，我们必须精心设置笑话，用适当的仪式来安抚文化机器。

This is how Siri was marketed from the beginning, as a culture machine so sophisticated it seemed to be magic: a computer that not only talks but understands. The commercials featured Samuel L. Jackson, Zooey Deschanel, and Martin Scorsese having actual conversations with their phones. Apple showed us that we could be human characters, crack wise, ask questions in our normal, idiosyncratic way, and that Siri would keep up, making its own jokes and understanding not just our words but the meaning behind them. The commercials hint at the kind of relationship that programmers flirt with through the “Easter eggs” they have left in Siri, such as the well-known trick of asking Siri where to dump a dead body (which offered a drily humorous rundown of local swamps, metal foundries, etc.—though Apple has since changed the response). These preprogrammed sparks illuminate the long-range goal to create not just intelligence but real personality. Yet these tricks come with a script that we must learn—for Siri to deliver each punch line we must carefully set up the joke, propitiating the culture machine with appropriate rituals.

该工具本身的界面为这些交互建立了具体的隐喻。一个巨大的麦克风按钮占据了屏幕，让人联想到一位电台主持人俯身收听电台节目，清晰地向空中吐字。当我们使用 Siri 时，这些吐字首先以模糊的数字符号、波浪线的形式出现，然后逐渐转化为对人类话语的引用转录。Siri 的回复随后以非引用的形式出现（有时只是口头表达，而不是记录在书面对话中）。这种设计强化了算法反馈回路，创造了一个系统，让我们学习 Siri 听到的内容，并逐渐训练我们用 Siri 理解的方式表达。或许是为了平衡这种简洁、典型的苹果设计风格，Siri 经常会扮演人类的角色，表达关切、讲笑话，并对用户的疑问发表意见。

The interface of the tool itself establishes concrete metaphors for these interactions. A large microphone button dominates the screen, suggesting the image of a radio host leaning over his station, enunciating clearly into the ether. As we use Siri those enunciations first appear as indeterminate digital symbols, waving lines that then resolve into a quoted transcription of the human utterance. Siri’s responses then appear as unquoted responses (and sometimes are only spoken, not inscribed in the written conversation). The design reinforces the algorithmic feedback loop, creating a system for teaching us what Siri hears and gradually training us to say things in ways that Siri understands. Perhaps to balance this sparse, typically Apple design scheme, Siri frequently takes on positions of human affect, expressing concern, making jokes, and opining about its user’s queries.

这些模拟人性的时刻是虚拟助手协作剧场的一部分：我们可以选择“捕捉”这些表演线索，并做出回应或忽略它们。当然，声音也持续不断地展现着性别。正如科技评论家安娜莉·纽维茨 (Annalee Newitz) 所言，数字助手中女性声音的主导地位显然与顺从有关：

These moments of simulated humanity are part of the collaborative theater of the virtual assistant: performative cues that we can choose to “catch” and respond to or ignore. And, of course, the voice offers a persistent performance of gender. As technology critic Annalee Newitz argues, the predominance of female voices in digital assistants clearly has something to do with submission:

可悲的事实是，这些数字助理比现代女性更像奴隶。他们不应该威胁你，也不应该与你平起平坐——他们应该服从命令，不予反抗。毕竟，理想的奴隶应该像你的母亲一样。她永远不会反抗，因为她爱你，无私地，永远地^。11

The sad truth is that these digital assistants are more like slaves than modern women. They are not supposed to threaten you, or become your equal—they are supposed to carry out orders without putting up a fight. The ideal slave, after all, would be like a mother to you. She would never rebel because she loves you, selflessly and forever.¹¹

性别的对话建构在人工智能的思想史中占有重要地位，我们稍后会回顾这个话题。但现在我们需要更深入地探索这一表现的后端。

The conversational construction of gender is prominent in the intellectual history of artificial intelligence, a topic we will return to. But for now we need to explore the back end of this performance in more depth.

在最黑暗的一面，这些数字助理被文化建构为具有感知能力的斯金纳箱，这些准智能系统必须做出回应，并激发用户同样的盲目崇拜。但当然，最终必须使用这些系统的是人类用户——不一定是Siri，而是那些评估我们搜索查询的互联系统，或者那些分配信用评分的算法。正如媒体学者西瓦·维迪亚纳坦等批评人士所指出的，不参与的代价是巨大的，但也难以确定，而算法文化的引力逐渐将参与的仪式、对特定计算祭坛的膜拜灌输进去^。12

At their darkest, these digital assistants are culturally constructed as sentient Skinner boxes, quasi-intelligent systems that must answer, and that inspire the same kind of slavish devotion in their users. But of course, it is the human users who in the end must use these systems—not necessarily Siri, but the interconnected systems that evaluate our search queries, for example, or the algorithms that assign credit scores. As critics like media scholar Siva Vaidhyanathan have pointed out, the price of nonparticipation is significant but also difficult to pin down, and the gravitational pull of algorithmic culture gradually inculcates the rituals of participation, of obeisance, to particular computational altars.¹²

本体论的魔力

The Magic of Ontology

Siri 的沟通方式对于文化理解至关重要，它认为智能助手是一种“有用的恶魔”——拥有特定且受限能力的实体——但这种成就的本质揭示了其更深层次的技术本质，超越了其对用户的实用性。Siri 作为文化机器的有效性，关键在于它实现了对问题进行快速、局部响应的最低可行阈值。Siri 解读现实世界命令的能力取决于两个关键因素：自然语言处理 (NLP) 和语义解读。任何尝试过在没有数据连接的情况下使用 Siri 的用户都知道，如果没有连接到 Apple 的服务器，该软件就无法运行。每次用户与 Siri 对话时，声音文件都会被发送到数据中心进行分析和存储，这是领先语音技术公司 Nuance 的一项服务。13^算法语音分析的重大突破源于放弃深层语言结构（即彻底映射语法和语义的努力），转而将语音视为一项统计和概率挑战。¹⁴给定这个音频信号，哪些文本字符串最有可能与每个单词相关联？哪个单词最有可能跟在另一个单词后面？用训练集或使用该服务的人类测试结果，并利用反馈逐步改进结果。这是一种经典的计算实用主义解决问题的方法，它依靠反复试验，在语言的泥沼中绘制出一条有效的可计算性路径，将口语视为任何其他复杂系统。

The call and response of Siri’s communication is central to the cultural understanding of intelligent assistants as a kind of useful demon—entities with specific, constrained abilities—but the nature of this achievement unveils a deeper technical being that exists beyond its utility to users. The vital element in Siri’s effectiveness as a culture machine is the achievement of a minimum viable threshold for speedy, topical responses to questions. Siri’s ability to interpret real-world commands depends on two key factors: natural language processing (NLP) and semantic interpretation. As any user who has tried to use Siri without a data connection knows, the software cannot operate without a link to Apple’s servers. Each time a user speaks to Siri the sound file is sent to a data center for analysis and storage, a service of the leading speech technology company Nuance.¹³ The major breakthroughs in algorithmic speech analysis have come by abandoning deep linguistic structure—efforts to thoroughly map grammar and semantics—in favor of treating speech as a statistical, probabilistic challenge.¹⁴ Given this audio signal, what text strings are most likely associated with each word? Which word is most likely to follow another? Test the results against a training set, or humans using the service, and use feedback to gradually improve results. This is a classic computational pragmatist approach to a problem, charting an effective computability pathway through the morass of language by depending on trial and error, treating spoken language just like any other complex system.

从这个意义上说，Siri 既是一项聆听服务，也是一项应答服务。随着时间的推移，Siri 大概收集了数十亿条成功和失败的互动记录，为改进语音识别提供了宝贵的资源。15^苹果声称其保留的数据是匿名的，但这一政策不出所料地让隐私权倡导者感到不安。16^虽然我们享受的是个性化服务，但 Siri 实际上是一台单一的集体机器，在工程师的监督下从这数十亿个数据点中学习。与许多其他大数据算法机器一样，它依赖于一个深井，一个容纳人类注意力和输入的蓄水池，作为计算推理的信息库。通过对足够多的数据进行统计语音分析，工程师可以生成人类表达的完整使用图谱，通过接受语言不是一套有待发现的固定规律体系（通用语法），而是一系列可以映射到选定准确度的行为，从而创建一个近乎通用的语音模型。

In this sense Siri is as much a listening service as it is an answering one. Over time Siri has presumably collected billions of records of successful and unsuccessful interactions, providing a valuable resource in improving speech recognition.¹⁵ Apple claims the data it retains is anonymized, but this policy is unsurprisingly troubling to privacy advocates.¹⁶ While we get personalized service, Siri is effectively a single collective machine, learning from these billions of data points under the supervision of its engineers. Like so many other big data, algorithmic machines, it depends on a deep well, a cistern of human attention and input that serves as an informational reservoir for computational inference. By running statistical speech analysis over enough data, engineers can generate a complete usage map of human expression, creating a near-universal speech model by accepting language not as a fixed system of laws to be discovered (a universal grammar) but as a series of behaviors that can be mapped to a chosen degree of accuracy.

即使采用这种大数据方法，正确地转录语音音频仍然是一项极具挑战性的任务。Siri 能否在其他语言版本上推出取决于开发有效的转录模型，而方言的语调变化对于操着英国口音、苏格兰口音和南方口音的用户来说，无疑是一大难题。17^但Siri 计算后端基于云端的特性意味着，每一种语言上的细微差别都会逐渐在苹果的数据中积累一个统计特征，也就是一系列记录，其中某个特定的单词或短语总是被误解。这些互动片段构成了 Siri 的自适应学习曲线，仿佛成千上万的方言使用者都在被动地教育同一个虚拟幼儿正确的发音，而 Siri 界面则同时塑造着这些用户的语调变化，鼓励他们以特定的、机器可识别的方式发音，并创建一种人机协作的语言分布曲线。

Even with this big data approach, correctly transcribing spoken audio is an incredibly challenging task. The rollout of Siri in other languages depends on developing effective transcription models, and dialect inflections have proven nettlesome barriers for users with British accents, Scottish burrs, and southern drawls.¹⁷ But the cloud-based nature of Siri’s computational back end means that each linguistic nuance gradually accumulates a statistical signature in Apple’s data, a series of recordings where a particular word or phrase is being consistently misinterpreted. These fragments of interaction contribute to Siri’s adaptive learning curve, as if thousands or millions of dialect speakers are all passively educating the same virtual toddler in correct pronunciation, while the Siri interface at the same time shapes the inflections of those users, encouraging them to enunciate in specific, machine-legible ways and creating a kind of collaborative human–machine distribution curve of language.

这种概率方法还将这些系统从人类的另一个限制中解放出来：就像蹒跚学步的幼儿一样，Siri 无需完全掌握语法即可进行有效的对话。由于算法的 NLP 不会尝试将每个句子解析为主语、动词和宾语，而是识别关键词和短语，因此 Siri 通过思想而不是语法来探索世界。18^这些关键词映射到 Siri 在结构化知识“本体”中已知的概念。对于 Siri 而言，语言始终代表世界上的对象：关键词“餐厅”附带其他概念，例如营业时间、实际地址和价格范围。19^当用户询问“点心”、“便宜”和“步行”时，Siri 可以有效地将这些关键词转换为连贯的查询，并在不了解动名词的情况下提供临时答案。

This probabilistic approach also frees these systems from another human constraint: like a toddler, Siri makes effective conversation without having a full grasp of grammar. Because the algorithm’s NLP does not attempt to parse every sentence into its subjects, verbs, and objects, but rather identifies key words and phrases, Siri navigates the world through ideas, not syntax.¹⁸ These keywords are mapped to concepts that Siri already knows about in structured “ontologies” of knowledge. For Siri, language always represents objects in the world: the keyword “restaurant” has other concepts attached to it, like hours of operation, a physical address, and a price range.¹⁹ When the user asks about “dim sum,” “inexpensive,” and “walking,” Siri can effectively translate these keywords into a coherent query and offer a provisional response without knowing anything about gerunds.

知识本体的概念结构支撑了 Siri 和其他智能代理所能实现的“魔力”。与人类一样，系统可以分析上下文，从之前的交互中学习，并根据不可靠的数据生成合理的猜测。在 CALO 项目中取得本体论突破的 SRI 工程师们也认识到了其可扩展性的价值：要插入一个新的电影评论、地理信息或其他类型的数据库，人类只需花费相对较少的时间来绘制新的知识领域（例如，实际地址或导演姓名）。本体的优势在于，它们能够在截然不同且快速变化的信息类型（从股票收益到短信）之间建立清晰的关联。

The conceptual structure of the knowledge ontology underwrites the “magic” of what Siri and other intelligent agents can accomplish. Like a human, the system can analyze context, learn from previous interactions, and generate plausible guesses on unreliable data. The SRI engineers who made the ontological breakthrough during the CALO project recognized its value for scalability as well: to plug in a new database of movie reviews, geographical information, or what have you, humans need to spend a relatively small amount of time mapping out the new fields of knowledge (e.g., physical address or name of the director). The strength of ontologies is that they establish clear relationships between radically different, rapidly changing kinds of information, from stock returns to text messages.

随着用户与不同数据集的交互，Siri 可以不断验证这些联系，并沿着 Siri 已知知识定义的严格控制的轴线扩展其人类语音模型。这一过程依赖于 Siri 数百万训练有素的协作者，这些用户必须学会以正确的方式格式化他们的查询，并考虑到系统显性和隐性的局限性。这些互动，尤其是其中的断层线，使我们能够标记出 Siri 中语言和知识的实现方式与更广泛的文化生活之间的差距。该系统对语言文化机器的模仿不仅在转录和语法方面存在缺陷，而且在其知识本体的意识形态建构方面也存在缺陷。例如，2011 年，Siri 曾一度陷入丑闻，因为有人爆出，​​查询“我在哪里可以堕胎”会返回反堕胎危机妊娠中心的结果，而这些中心通常位于很远的地方，并且无法代表提供堕胎服务的当地计划生育诊所。²⁰统计语言模型将查询与更有可能在其网站上使用“堕胎”一词的反堕胎机构联系起来，未能理解该问题的政治色彩。Siri 知识本体的黑匣子结构混淆了系统因排除计划生育机构而犯下的分类错误。修复文化机器中的这一故障必然需要人为干预：在黑匣子的背后，工程师必须用例外和变通方法推翻基线统计模型。肯定有成千上万个这样的例外，特别是对于模仿人类情感的反应。Siri 及其各种同行提供了一种通用语言计算的愿景，但在实践中依赖于需要不断调整和监督的“有效”计算。

Siri can validate these connections over time as users interact with different datasets, extending its model of human speech along tightly controlled axes defined by what Siri already “knows.” The process depends on Siri’s millions of trained collaborators, the users who must learn to format their queries the right way and take the system’s explicit and implicit limitations into account. These interactions, and particularly their fault lines, allow us to mark the gap between the ways language and knowledge are implemented in Siri as compared to in broader cultural life. The system’s mimicry of the linguistic culture machine is imperfect not just in transcription and grammar, but in the ideological construction of its knowledge ontologies. For example, in 2011 Siri was briefly engulfed in scandal when it emerged that queries for “where can I get an abortion” would return results to anit-abortion crisis pregnancy centers, often far away, and fail to represent local family planning clinics that provided abortion services.²⁰ The statistical language model linked the query to anti-abortion facilities that were far more likely to use the word “abortion” on their websites, failing to grasp the politically charged nature of the question. The black box structure of Siri’s knowledge ontology obfuscated the category error the system made by excluding Planned Parenthood facilities. Fixing this glitch in the culture machine necessarily involves human intervention: behind the facade of the black box, engineers had to overrule baseline statistical models with exceptions and workarounds. There must be thousands of such exceptions, particularly for responses that mimic human affect. Siri and its various counterparts offer a vision of universal language computation, but in practice depend on an “effective” computation that requires constant tweaking and oversight.

正如堕胎案例所示，在算法自训练智能助手的前提与实践中神秘的黑匣子功能之间存在着巨大的抽象鸿沟。这台文化机器及其所包含的知识本体，对我们与它之间所能建立的智力关系类型施加了深刻的限制，因为虽然Siri可以向我们学习（例如，通过汇总数十亿条语音记录进行分析），但我们几乎无法直接教它任何东西。本体是Siri的秘密武器，是维系整个运作的神经突触，它们是根据苹果及其合作伙伴的商业逻辑、法律协议和许可方案构建的。这可能会导致一些失败，例如Siri在回答“婴儿从哪里来”这个问题时，会提供该地区的“婴儿用品店”列表。21^但这也会导致一些更难察觉的缺失和排除。隐形本体不会暴露其接缝和边缘，而恰恰是为了抵制这类质疑而设计的。算法系统通过对数百万次对话进行非个人化的汇总来积累意义；而 Siri 的人类用户则根据数量少得多的深度个人互动来赋予意义。

As the abortion example demonstrates, there is a gulf of abstraction between the premise of the algorithmically self-training intelligent assistant, on the one hand, and the mysterious functionality of the black box in practice. This culture machine and the knowledge ontologies it contains impose profound limits on the kinds of intellectual relationships we can have with it, because while Siri can learn from us (by aggregating billions of voice recordings for analysis, for example), we can directly teach it almost nothing. The ontologies are Siri’s secret sauce, the synapses that hold the entire operation together, and they are constructed according to the business logic, legal agreements, and licensing schemes of Apple and its partners. This can lead to failures like Siri responding to the question “where do babies come from” by offering a list of “baby stores” in the area.²¹ But it also leads to absences and exclusions that can be harder to detect. An invisible ontology does not reveal its seams and edges, but rather is designed to resist exactly that kind of questioning. The algorithmic system accrues meaning by the impersonal aggregation of millions of conversations; Siri’s human users assign meaning on the basis of a much smaller set of deeply personal interactions.

这两种对算法的理解之间偶尔会发生的惊人冲突激怒了许多果粉，因为他们认为这违反了他们与苹果达成的默契：为符合广告宣传的高质量产品支付溢价。22^通过将技术直接映射到文化上，苹果无意中揭示了将一个仍处于测试阶段的产品连接在一起的纽带，从而在算法的计算主义承诺与支撑它的真正文化机器之间创造了另一种紧张的关系。由于 Siri 与用户如此紧密地联系在一起，处理提醒、消息和口头查询，从而深入了解个人生活，因此失望更加强烈。Siri 的吸引力，以及之前的 CALO 的吸引力，已经超越了单纯的助手，而更像是一个比你更了解你生活的贴心伴侣。Siri 是通往完美自我认知之路上的一个驿站。

The occasionally spectacular clashes between these two understandings of the algorithm angered many Apple fans because they saw them as a violation of the implicit bargain they have made with the company: to pay a premium for high quality products that live up to their advertising.²² By mapping technology directly onto culture, Apple inadvertently revealed the stitching holding together a product still in beta, creating another fraught relationship between an algorithm’s computationalist promise and the real culture machine underpinning it. The disappointment was more acute because Siri operates so intimately with its users, handling reminders, messages, and verbal queries that provide a deep index into personal life. The appeal of Siri, and indeed of CALO before it, goes beyond just an assistant to something like an ever-attentive companion who knows your life better than you do. Siri is a way station on the path to the quest for perfect self-knowledge.

星际迷航计算机

The Star Trek Computer

如果说Siri成为“智能助手”的追求雄心勃勃，那么谷歌在算法文化前沿的努力则更为深远。该公司已经涉足了大量的在线活动：2013年，谷歌直接产生了美国25%的互联网流量，每天约有60%的互联网设备与谷歌服务器交换数据。23从Gmail到YouTube，谷歌提供的众多免费工具和服务，让数亿人（以及他们产生的数据）与谷歌建立了联系，这种联系可能像在公共图书馆终端机上进行一次简单的搜索一样肤浅，也可能像长达^十年的个人档案和信件纠葛一样深厚。正如Siva Vaidhyanathan在2011年评论谷歌时所写：“从来没有一家公司明确地雄心勃勃地要在全球范围内——实际上是在全世界范围内——将个人思想与信息连接起来。” ²⁴

If Siri’s quest to become an “intelligent assistant” seems ambitious, Google’s efforts at the cutting edge of algorithmic culture are even more sweeping. Already the company is involved in a huge portion of activity online: in 2013, Google generated 25 percent of all Internet traffic in the United States directly, and roughly 60 percent of all devices on the Internet exchanged data with Google servers on any given day.²³ Its proliferation of free tools and services, from Gmail to YouTube, have brought hundreds of millions of people (and the data they generate) into relationships with the company that might be as superficial as a single search at a public library terminal or as deep as decade-long entanglements of personal archives and correspondence. As Siva Vaidhyanathan wrote of Google in 2011: “there has never been a company with explicit ambitions to connect individual minds with information on a global—in fact universal—scale.”²⁴

谷歌的庞大规模及其与算法文化核心结构的深度交织也激发了其全球扩张的雄心。该公司的X实验室致力于研究“登月计划”，旨在为当前挑战提供彻底的解决方案或指数级改进。他们是谷歌眼镜、自动驾驶汽车和“气球计划”（Project Loon，旨在通过高空气球为偏远地区提供互联网服务）等高风险项目背后的智力力量。实验室“登月计划”负责人阿斯特罗·泰勒（Astro Teller）鼓励快速原型设计和早期失败点的文化，以尝试新想法。25自²⁰¹⁴年收购人工智能研究机构DeepMind以来，谷歌还发布了一系列关于机器学习进展的惊人公告。但尽管有这些创新，该公司（及其新的控股公司Alphabet）的几乎所有收入都来自广告，这不仅为其带来了利润，也为其提供了积极探索新市场和新想法的动力。

Google’s immensity and deep imbrication in the core structures of algorithmic culture have also fueled global ambitions. The company’s X Lab dedicates itself entirely to considering “moonshot” ideas that offer radical solutions or exponential improvements to current challenges, and they are the intellectual force behind high-risk ventures such as Google Glass, the self-driving car, and Project Loon, an effort to deliver Internet service to remote areas via high-altitude balloons. Astro Teller, the lab’s captain of moonshots, has encouraged a culture of rapid prototyping and early failure points to try out new ideas.²⁵ Since the company acquired the artificial intelligence research group DeepMind in 2014, Google has also made a string of breathtaking announcements about advances in machine learning. But for all of these innovations, the company (and its new holding corporation, Alphabet) makes essentially all of its money from advertising, giving it both the profits and the motivation to aggressively explore new markets and ideas.

当你将谷歌视为数字文化的仲裁者时，它的雄心壮志背后有着广泛而合理的商业逻辑。搜索栏的简洁实用性，以及Gmail、YouTube和其他重要服务的界面，掩盖了其深厚的基础架构，其最终目的是构建一个信息宇宙的一致模型。2012年，谷歌宣布将整合59种不同服务的用户信息，为全面理解数字文化领域中人类的行为奠定基础。26^鉴于谷歌众多服务提供的广泛互联，该公司如今拥有足够的数据来开发极其复杂的用户意图模型。

When you look at Google as an arbiter of digital culture, its ambitions have a vast but plausible business rationale. The spare utility of the search bar or the interfaces for Gmail, YouTube, and other essential services mask a deep infrastructure designed, ultimately, to construct a consilient model of the informational universe. In 2012, Google announced that it would integrate user information across fifty-nine different services, laying the groundwork for a holistic understanding of human behavior across many spheres of digital culture.²⁶ Given the pervasive interconnections that Google’s many services offer, the company now has the data to develop startlingly sophisticated models of user intentions.

这种技术官僚野心的态度也许在公司董事长、《新数字时代：重塑人类、国家和商业的未来》一书的作者埃里克·施密特身上得到了最好的体现。2010 年，施密特向《华尔街日报》表示，“我实际上认为大多数人不希望谷歌回答他们的问题。他们希望谷歌告诉他们下一步应该做什么。” ²⁷这句简单的评论表明了谷歌角色的深刻转变，标志着他们从一家在算法时代掌握信息访问权的公司转变为一家想要打造星际迷航计算机的公司。这是一个阶段性的变化，从以世界上最伟大的搜索引擎的形式积累技术知识，到具有自己的机构和影响结构的进化技术形式。

This attitude of technocratic ambition is perhaps best reflected in the company’s chairman Eric Schmidt, author of The New Digital Age: Reshaping the Future of People, Nations and Business. In 2010, Schmidt argued to the Wall Street Journal, “I actually think most people don’t want Google to answer their questions. They want Google to tell them what they should be doing next.”²⁷ This simple comment articulates a profound shift in Google’s role, one that marks their transition from the company that mastered information access in the age of the algorithm to the company that wants to build the Star Trek computer. It is a phase change from the accumulation of technical knowledge, in the form of the world’s greatest search engine, to an evolved form of technical being with its own agency and structures of influence.

这一演变是该公司长期以来精心考量的目标，核心搜索工具的工程师和高管也反复强调这一点。例如，搜索产品管理总监塔玛·耶霍舒亚（Tamar Yehoshua）曾说过：“我们的愿景是打造出一台星际迷航般的电脑。你可以跟它说话——它能理解你，也能和你对话。” ²⁸

This evolution is a longstanding and closely examined goal for the company, repeatedly articulated by engineers and executives in charge of its core search tools. To cite just one example, from Tamar Yehoshua, director of product management for search: “Our vision is the Star Trek computer. You can talk to it—it understands you, and it can have a conversation with you.”²⁸

星际迷航计算机到底是什么？令人惊讶的是，在浩瀚的星际迷航宇宙的大框架内，这个提法相当肤浅。从《星际迷航：下一代》开始，图书馆计算机访问/检索系统 (LCARS) 以各种形式出现在企业号和星际舰队联盟的其他舰船上。LCARS 主要因其未来主义的视觉设计而引人注目（最初由布​​景设计师 Michael Okuda 设计，为了节省昂贵的闪烁灯光），但在叙事中很少扮演重要角色。当然，除非出现问题，计算机被某些恶意敌人控制（例如“进化”一集中的纳米机器人）。29^与该剧叙事背景的其他元素一样，企业号的会说话的计算机本应平淡无奇且高效。30 《星际迷航》的对话计算机有其局限性，它会滑稽地误解请求，偶尔还会激发我们许多人现在在语音驱动的算法系统中使用的那种生硬的“关键词语” ^。在其巅峰时期，它作为数据科学的一种自然语言界面，在各种信息中寻找模式并进行分析。31^最重要的是，它提出了一种简单的无摩擦语音交互的理想：谷歌所说的星际迷航计算机，以及 LCARS 在该剧情节中简单而有效的作用，就是对口头命令和查询做出有用的回应。

What exactly is the Star Trek computer? Surprisingly, the reference is fairly shallow within the broader framework of the vast Star Trek universe. The Library Computer Access/Retrieval System (LCARS) appears in various guises on the Enterprise and other ships of the Star Fleet Federation beginning with Star Trek: The Next Generation. Remarkable mostly for its futuristic visual design (originally created by set designer Michael Okuda to save money on expensive blinking lights), LCARS rarely plays a significant role in the narrative. Unless, of course, something goes wrong and the computer has been possessed by some malicious enemy (like the nanites in the episode “Evolution”).²⁹ Like other elements of the diegetic background of the show, the Enterprise’s talking computer was meant to be unremarkable and efficient.³⁰ The conversational computer of Star Trek had its limits, comically misunderstanding requests and occasionally inspiring the kind of stilted “keywordese” many of us now use with voice-driven algorithmic systems. At its peak, it served as a kind of natural language interface for data science, seeking patterns in various kinds of information and presenting analysis.³¹ Most important, it presented a simple ideal of frictionless vocal interaction: what Google appears to mean by the Star Trek computer, and what LCARS does simply and effectively for the show’s plotting, is respond usefully to verbal commands and queries.

正如《星际迷航》中人物向LCARS提出的问题所表明的那样，一台会说话的计算机本身的作用远不止呼叫人员和订茶。CALO系统已经证明，语音驱动的界面只是更为复杂的文化机器的一部分。“图书馆计算机访问/检索系统”这个名称暗示了这一点：这台计算机必须是所有人类知识的索引，一个拥有智能数字代理的图书馆，可以解析任何语音查询并返回相关信息。LCARS是《星际迷航》宇宙的文化机器，是星际联邦知识本体的替代者。该剧片头字幕中的“舰长誓言”承诺，美国企业号航空母舰将成为一种不同类型的搜索引擎，一种造价昂贵的工具，“去探索陌生的新世界，寻找新的生命和新的文明，勇敢地前往前人未至之地”。即使在20世纪60年代，人们也已经意识到，此类活动将产生海量数据，这些信息需要被收集、关联并组织成一个一致的整体结构，以便LCARS系统能够搜索这些信息，从而有效地回答有关行星大气或飞船日志中条目的查询。《星际迷航》中的计算机是一个界面，它实现了一个更广泛的目标：以一致的方式组织信息：一个值得探索宇宙知识的基础设施。

As the queries Star Trek characters pose to LCARS indicate, a talking computer on its own is not useful for much more than paging people and ordering tea. The CALO system already demonstrated that a speech-driven interface is only one part of a much more complex culture machine. The name Library Computer Access/Retrieval System hints at it: this computer must be an index to all human knowledge, a library with an intelligent digital agent which can parse any spoken query and deliver relevant information back. LCARS is a culture machine for the Star Trek universe, a stand-in for the knowledge ontology of the United Federation of Planets. The “captain’s oath” in the opening credits of the show pledges the U.S.S. Enterprise to be a different sort of search engine, a massively expensive instrument “to explore strange new worlds, to seek out new life and new civilizations, to boldly go where no one has gone before.” Even in the 1960s it was obvious that such activities would generate vast troves of data, information that would need to be collected, correlated, and organized into a consilient overall structure, something that LCARS could search in order to effectively answer queries about a planet’s atmosphere or entries in the ship’s log. The Star Trek computer is the interface for a much broader goal of organizing information in a consistent way: an infrastructure worthy of the quest for universal knowledge.

启蒙算法

The Algorithm of the Enlightenment

在沉思的时刻，《星际迷航》的船长们常常提醒我们，他们受好奇心驱使的使命背后蕴藏着怎样的思想史，并将他们的冒险经历置于更古老的普遍知识探索的背景下。从我们最早的起源神话开始，这个梦想就一直萦绕在人类心头，但过去五百年来主导技术社会文化的这种形成性变体，源于欧洲启蒙运动时期科学理性主义的胜利。如果谷歌试图构建一个普遍的知识体系，并试图以商业形式实现这一目标，那么这个项目就有一个重要的先例。这正是丹尼斯·狄德罗、让·达朗贝尔和其他百科全书编纂者在十八世纪中叶法国所追求的目标。正如文化历史学家兼图书管理员罗伯特·达恩顿在《启蒙运动的事业》一书中所描述的，第一部百科全书（1751年）的创作及其随后的普及是一项极其昂贵的投机性事业。 “ 《百科全书》出版工作的规模之大，充分表明了百科全书的重要性，因为正如它的朋友和敌人一致认为的那样，这本书代表着某种比其本身更宏大的东西，一场运动，一种‘主义’。它已经成为启蒙运动的象征。” ³²在谷歌以广泛而深刻的方式体现数字知识的状态（Siva Vaidyanathan 称之为“万物谷歌化”）之前，《百科全书》就承担了同样激进的项目。

In their contemplative moments, Star Trek captains often remind us of the intellectual history that underpins their curiosity-driven mission, framing their adventures in the context of a much older quest of universal knowledge. That dream has haunted humanity from our earliest origin myths, but the formative variation that has dominated technosocial culture for the past five hundred years springs from the triumph of scientific rationalism in the European Enlightenment. If Google is trying to assemble a universal structure of knowledge, and attempting to do so as a commercial enterprise, the project has an important precedent. This was precisely the aim that Denis Diderot, Jean D’Alembert, and the other encyclopédistes were striving for in mid-eighteenth-century France. As cultural historian and librarian Robert Darnton describes in The Business of the Enlightenment, the creation of the first Encyclopédie (1751) and its subsequent popularization were tremendously expensive, speculative enterprises. “The sheer scale of the Encyclopédie publishing effort suggests the importance of Encyclopedism, for as its friends and enemies agreed, the book stood for something even larger than itself, a movement, an ‘ism.’ It had come to embody the Enlightenment.”³² Before Google came to embody, in expansive and profound ways, the state of digital knowledge (what Siva Vaidyanathan has called the Googleization of Everything), the Encyclopédie took on the same radical project.

《百科全书》在当时是一本危险的书，因为它呈现了一种整体的、世俗的人类知识观——不仅仅是从A到Z的一系列主题，而是计算机科学家所称的“信息的新本体论”。书中的插页如同一份视觉宣言，宣告记忆（包括神圣和宗教的记忆）、理性和想象力之间的平等（图 2.2）。本体论代表了关于知识等级制度的意识形态立场，将主题组织成既有谱系又有决定性的树状图。狄德罗和达朗贝尔提出的本体论自发表以来一直受到密切讨论，正是因为它为知识世界提供了一种新的模型，即“心灵的哲学史”。33^这样，百科全书本身就是一个技术存在，一个有自身能动形式的构造，它塑造了人类和技术事件的进程，从启蒙运动理想的逐渐全球传播到更为直接的法国大革命。

The Encyclopédie was a dangerous book for its time because it presented a holistic, secular view of human knowledge—not just a list of topics from A to Z, but what computer scientists would call a new ontology of information. An insert to the volume served as a visual manifesto, a declaration of equality between memory (including the sacred and religious), reason, and imagination (figure 2.2). Ontologies present an ideological position about the hierarchy of knowledge, organizing subjects into trees that are both genealogical and determinative. The ontology Diderot and D’Alembert presented has been closely discussed more or less continuously since its publication precisely because it presented a new model for the intellectual universe, a “philosophic history of the mind.”³³ In this way, the Encyclopédie was its own technical being, a construct that had its own form of agency and that shaped the course of human and technological events, from the gradual, global diffusion of the Enlightenment ideal to the far more immediate French Revolution.

百科全书是一个过程，一台旨在推动世界发生特定形式变革的文化机器。达朗贝尔是第一卷著名的《初步论述》的作者，他不遗余力地承认该项目受益于勒内·笛卡尔，后者的《方法论》（1637）在其标题中预示了我们或许可以称之为启蒙运动的算法。怀疑主义、理性主义、归纳推理以及公众对知识的批判性评价都是这种方法的要素，一个世纪后，这种方法在狄德罗和达朗贝尔的推动下达到了转折点。正如达朗贝尔在序言中所说，《百科全书》是一种改变知识进步进程的方法，一个过程：

The encyclopedia was a process, a culture machine intended to work very specific forms of change in the world. D’Alembert, author of the first volume’s celebrated “Preliminary Discourse,” takes pains to acknowledge the project’s debts to René Descartes, whose Discourse on the Method (1637) prefigures in its very title what we might call the algorithm of the Enlightenment. Skepticism, rationalism, inductive reasoning, and the public, critical evaluation of knowledge are all elements of this method, which reached a tipping point a century later with Diderot and D’Alembert. As D’Alembert’s introduction argues, The Encyclopédie is a method, a process, for changing the course of intellectual progress:

在未来的科学和人文学科成果中，将很容易区分发明者从自身资源中汲取的成果和从前人那里借用的内容。作品将得到准确的评价，那些渴望名声、缺乏天赋、厚颜无耻地将旧系统当作新思想发表的人，很快就会被揭穿^。34

In the future productions of the sciences and liberal arts it will be easy to distinguish between what the inventors drew from their own resources and what they borrowed from their predecessors. Works will be accurately evaluated, and those men who are eager for reputation and devoid of genius, and who brazenly publish old systems as if they were new ideas, will soon be unmasked.³⁴

该算法的实施规模与谷歌搜索不相上下。正如书籍史学家所指出的，《百科全书》的投机性冒险不仅仅涉及一本书的出版，还涉及一种新型媒介的发明。达恩顿对破布从后门离开资产阶级家庭，最终又作为新百科全书的页面重新进入正文的精彩叙述，只是创作这本极其昂贵的书籍所必需的材料生产过程之一。现代百科全书的诞生，包括交叉引用、按字母顺序排列的主题和反思性的知识本体论，需要一系列新的编辑、制作和商业系统来将这项发明传播到世界各地。

This algorithm was implemented on just as grand a scale as Google Search. As historians of the book have pointed out, the speculative venture of the Encyclopédie did not merely involve the publication of a book but the invention of a new kind of medium. Darnton’s lovely narration of rags leaving bourgeois homes from the back door, only to re-enter the front as pages in the new encyclopedia, is just one of the processes of material production necessary for the creation of this extraordinarily expensive book. The birth of the modern encyclopedia, with cross-references, alphabetized topics, and a reflexive ontology of knowledge, required a slate of new editorial, production, and commercial systems for the transmission of this invention across the world.

《百科全书》的革命性不仅在于它将上帝和宗教从知识的首要地位中抽离出来，还在于其方法和本体论要求持续改进。“引用事实、比较实验、阐述方法，只是为了激发天才开辟未知的道路，并利用伟人终结职业生涯的地方作为第一步，继续探索新的发现。” ³⁵作为一个激进的理念，《百科全书》取得了如此彻底的成功，以至于很难想象其他的可能性。它的导言和“百科全书”条目勾勒出一幅建立在百科全书算法可完善性基础上的理性进步的长远愿景。文化机器不断学习，通过“文人和艺术家团体”的协同合作，通过编辑、交叉引用、印刷、出版和辩论等智力机制来纠正其早期的缺陷。³⁶启蒙运动的算法，即通过对事实进行排序、测试和比较来寻求完美的知识，是谷歌星际迷航计算机的核心。

The Encyclopédie was revolutionary not merely for its evacuation of God and religion from intellectual primacy, but also because its method and ontology called for continuous improvement. “Facts are cited, experiments compared, and methods elaborated only in order to excite genius to open unknown routes, and to advance onward to new discoveries, using the place where great men have ended their careers as the first step.”³⁵ As a radical idea, the Encyclopédie succeeded so completely that it is difficult to imagine other possibilities. Its introduction and the entry “Encyclopedia” sketch out a vision for the long arc of rational progress, built on the perfectibility of the encyclopedic algorithm. The culture machine learns, correcting its early flaws through the concerted collaboration of a “society of men of letters and artists” who work through the intellectual mechanisms of editing, cross-referencing, printing, publicizing, and debate.³⁶ The algorithm of the Enlightenment, the quest for perfect knowledge through the ordering, testing, and comparison of facts, is at the heart of Google’s Star Trek computer.

双重知识探索

The Twin Quests for Knowledge

谷歌星际迷航计算机的后端或深层结构是谷歌业务的核心：索引算法、数据存储和信息管理工具，这些使其成为全球领先的搜索引擎。我将在第五章讨论该架构的众所周知的基础，但在这里，我想谈谈谷歌将其野心进一步拓展到百科全书领域的一些粗糙之处：一个名为“知识图谱”（KnowledgeGraph）的全新本体论项目。Siri 依赖于一组相对较小的精选数据分类法（例如，来自 OpenTable 的数据可能包括餐厅名称、电话号码、日历可用性等等），而“知识图谱”则试图在其搜索爬虫程序可用的全部数据上创建类似的映射。“知识图谱”是一个开放的本体论，它从维基百科等主要由人工编辑的“受控”来源获取信息，也从谷歌扫描的所有网页的非结构化数据中获取信息^。37

The back end or deep structure of Google’s Star Trek computer is the core of Google’s business: the indexing algorithms, data storage, and information management tools that have made it the world’s leading searching engine. I address the well-known foundations of that architecture in chapter 5, but here I want to talk about the rough edges where Google is expanding its ambitions deeper into Encyclopédie territory: a sweeping new ontological project called KnowledgeGraph. Where Siri depends on a relatively small set of curated data taxonomies (e.g., data from OpenTable might include restaurant names, phone numbers, calendar availabilities, and so on), KnowledgeGraph attempts to create similar mappings on the full swath of data available to Google from its search crawlers. KnowledgeGraph is an open ontology, drawing information from “controlled” sources like Wikipedia that are primarily human-edited, but also from the unstructured data of all the web pages Google scans.³⁷

知识图谱的方法呼应了狄德罗及其前辈，特别是莱布尼茨“万有数学”的普遍主义梦想。它已经包含了数亿个实体，包括多种语言的人物、事物、文化作品和地点，所有这些都通过数十亿种关系相互关联。38^这种态度反映在谷歌工程师用来描述这一努力的语言中。高级副总裁兼首席搜索工程师阿米特·辛格哈尔 (Amit Singhal) 将搜索的未来想象成一种回答任何问题的工具，就像《星际迷航》一样：“如果我们谷歌能够回答真正的问题，我们的用户就会变得更加博学，他们在寻求知识的过程中也会更加满足。” ³⁹就像狄德罗的原著一样，这个百科全书资源不是静态形式，而是一个改善用户生活的过程。

The KnowledgeGraph approach echoes both Diderot and his antecedents, particularly the universalist dream of Leibniz’s mathesis universalis. It already contains hundreds of millions of entities that include people, things, cultural works, and places in multiple languages, all interlinked by billions of relationships.³⁸ That attitude is reflected in the language Google engineers use to describe the effort. Amit Singhal, senior vice president and chief search engineer, imagines the future of search as a tool to answer any question, just like Star Trek: “Genuine questions that, if we, Google, can answer, our users would become more knowledgeable and they would be more satisfied in their quest for knowledge.”³⁹ Like Diderot’s original, this encyclopedic resource is not a static form but a process for improving the lives of its users.

通过想象我们都在寻求知识来展望搜索的未来，一方面将我们与计算之间浪漫的关系理想与一套独特的理性主义、经验主义的算法工具相结合，以实现这一目标。知识图谱 (KnowledgeGraph) 旨在对“有效计算”做出一个雄心勃勃的定义，涵盖公共网络上所有可访问的事实。与 Siri 的事实数据库一样，知识图谱不仅编码信息，还编码关系，这是一种对最终用户来说基本上不可见的元结构，允许根据地理位置、日期或高度等指标进行比较。当然，与百科全书不同，谷歌的知识树仍然是不可见的，就像一套隐藏在黑匣子里的商业秘密：解决歧义和相互冲突的事实的方法默认隐藏在数据库的逻辑中。随着时间的推移，该公司的本体论将不断扩展，但只会部分地展现出来，甚至可能向每个用户展示其不同的一面，即使它掌握着一个不断增长的“客观”事实和关系网络。

To approach the future of search by imagining that we are all on a quest for knowledge combines the romantic ideal of our relationship with computation, on the one hand, and a distinctly rationalist, empirical set of algorithmic tools for realizing that goal on the other. KnowledgeGraph aims for an ambitious definition of “effective computing” that encompasses every fact accessible on the public web. Like Siri’s database of facts, KnowledgeGraph codes not only information but relationships, a metastructure that is largely invisible to the end user, allowing comparisons along metrics like geographical location, date, or height. Of course, unlike the Encyclopédie, Google’s tree of knowledge remains invisible, a set of trade secrets hidden in black boxes: the ways that ambiguities and conflicting truths are resolved remain hidden by default within the logic of the database. Over time the company’s ontology will expand but will reveal itself only in parts, perhaps even showing a different side of itself to each user, even as it masters a growing network of “objective” facts and relationships.

对自适应搜索结果的接受标志着我们脱离了启蒙运动的普遍知识算法。谷歌这一探索的动机框架不仅仅是拥有知识，而是通过查询来呈现知识：探索的智力模式与目的地本身同样重要。2013 年，辛格哈尔发表了题为“我们所知的搜索的终结”的主题演讲，他用一个简单的清单描述了公司的抱负：谷歌的算法需要回答、交谈和预测。40^这些代表了一种智力需求层次，从仅仅积累知识转变为一种全新的意向性形式。如果我们将知识图谱视为“回答”的最新技术，那么 Siri 和谷歌在自然语言语音处理方面的对应技术则是“交谈”的巅峰。但金字塔的顶端“预测”则完全是另一个问题。总而言之，这三个术语巧妙地重塑了百科全书的本体论：回答就是掌握历史；交谈就是掌握理性；预测需要想象力（图2.2）。但将预测交给计算算法，却挑战了追求完美知识的更广泛基础。本体论之树的根基不再建立在“文人和艺术家的社会”中，而是由工程师（仍然主要由男性组成）和机器组成的私人企业结构。

This embrace of adaptive search results marks our departure from the Enlightenment algorithm of universal knowledge. Google’s motivating framework for this quest is not merely to have the knowledge, but to present it via query: the intellectual mode of the quest is just as important as the destination itself. In 2013, Singhal delivered a keynote titled “The End of Search as We Know It” where he described the company’s ambitions with a simple list: Google’s algorithms need to answer, converse, and anticipate.⁴⁰ These represent an intellectual hierarchy of needs, moving from merely accumulating knowledge to an entirely new form of intentionality. If we think of KnowledgeGraph as the state of the art for “answering,” Siri and Google’s counterpart in natural language speech processing are the current pinnacle for “conversing.” But the top of the pyramid, “anticipating,” is another question entirely. Taken together, the three terms neatly reframe the ontology of the Encyclopédie: to answer is to master history; to converse is to master reason; to anticipate requires imagination (figure 2.2). But putting anticipation in the hands of computational algorithms challenges the broader foundations of the quest for perfect knowledge. The root of the ontological tree is no longer founded in a “society of men of letters and artists” but instead in a private, corporate structure made up of engineers (still mostly men) and machines.

当我们将启蒙运动的算法个性化时会发生什么？正因如此，谷歌雄心壮志的最后一个要素也是文化层面最复杂的要素：预见需要构建一台不仅能理解公共文化，还能理解私人意图的机器。与Siri一样，谷歌不需要完整的语法结构或社会习俗地图，但它需要理解人类思维的概率运作。谷歌清楚地意识到，我们真正的问题无法由百科全书来解答，无论它多么普及。施密特在2010年就秉持了这一立场：“[人们]希望谷歌告诉他们下一步应该做什么。” 对谷歌而言，这种探索的逻辑意味着，它努力达到普遍知识的境界，也需要实现完美的自我认知，以类似丘奇-图灵论题的、有效可计算的方式预测用户的需求和疑问。

What happens when we personalize the algorithm of the Enlightenment? This is why the final element of Google’s ambition is also the most culturally complex: anticipation requires building a machine that understands not just public culture but private intention. Like Siri, Google does not need a full map of grammatical structures or social mores, but it needs to understand the probabilistic functioning of the human mind. Google is clearly aware that the real questions we have cannot be answered by an encyclopedia, no matter how universal. This is the position from which Schmidt argued in 2010 that “[people] want Google to tell them what they should be doing next.” For Google, the logic of the quest means that its effort to reach a state of universal knowledge also requires the achievement of perfect self-knowledge, of anticipating its users’ needs and queries in a Church–Turing thesis–like, effectively computable way.

谷歌最接近预期的，是其根据个人用户情境构建其海量数据资源的系统。Google Now 收集用户使用的每一项谷歌服务的个人数据，其口号是“在恰当的时间提供恰当的信息”。这项服务虽然仍然有限，但却预示着算法与人类之间一种全新的共生关系。用户会在谷歌认为最相关的时刻收到通知和信息，例如，它会提醒用户上车以便准时参加下次会议，并建议最佳出行路线。航班预订、订单跟踪信息和新闻条目会根据系统对用户在特定时刻可能想要了解的内容的猜测而弹出。

The closest Google has come to anticipation is the system that frames its vast data resources through the context of the individual user. Google Now draws on personal data from every Google service an individual uses, under the tagline, “Just the right information at just the right time.” The service, while still limited, promises an entirely new kind of symbiotic relationship between algorithm and human. The user receives notifications and pieces of information at the moment Google thinks they may be most relevant, for example nudging a user to get in the car to arrive on time for his next meeting and suggesting optimal travel routes. Flight reservations, order tracking information, and news items pop up based on the system’s guesses about what the user might want to know at a particular moment.

“预期”是对上述两种互动模式的大胆突破，它甚至可能超越了《星际迷航》中的计算机本身——它表明，对于百科全书编纂者来说，想象力是创意和文学艺术的领域。目前，Google Now 会建议最佳的出行时间和路线。不难想象，未来系统会回答更广泛的“我们下一步应该做什么”的问题，尤其是应该读什么、看什么、买什么，以及我们应该去哪里、应该见谁，以及间接地，我们应该思考和做什么。谷歌几乎无处不在的网络，它与无数文化系统的交织，不仅赋能，而且有效地定义了数十亿人的特定文化活动领域，这使得它不仅仅是一种建议服务，甚至不仅仅是一种复杂的广告形式。谷歌文化机器正在绘制一张地图，有时甚至有可能抢走原有的版图。

“Anticipation” is a bold departure from these other two modes of interaction, and it arguably eclipses the Star Trek computer itself—it is telling that for the encyclopédistes imagination was the realm of the creative and literary arts. Currently Google Now will suggest optimal timings and pathways through the world. It is not hard to imagine a future where the system answers a much broader set of “what we should be doing next” questions, especially what to read, watch, and buy, but also where we should be, who we should see, and, indirectly, what we should think and do. Google’s near-omnipresence online, its imbrication in countless cultural systems that do not merely enable but effectively define certain cultural fields of play for billions of people, make this more than just a suggestion service or even a sophisticated form of advertising. The Google culture machine is assembling a map that at times threatens to upstage the territory.

星际迷航中的计算机与谷歌正在构建的计算机之间的对比，展现了启蒙算法与新兴“预期”逻辑之间的巨大差距。企业号是对狄德罗和达朗贝尔理论的直接升级，是人类探求知识的神奇工具。计算机或多或少地通过开门、回答参考问题、运行生命维持系统以及保持照明等功能来预测船长和船员的需求——仅此而已。但对于谷歌的“搜索终结”而言，对知识的追求本身就是工程目标：构建一个系统，为我们完成将万物连接在一起的艰巨文化工作。

The contrast between the Star Trek computer and what Google is building illustrates the gulf between the Enlightenment algorithm and the emerging logic of “anticipation.” The Enterprise is a straightforward update to Diderot and D’Alembert, a marvelous instrument for a very human quest for knowledge. The computer anticipates the needs of the captain and crew more or less by opening doors, answering reference queries, running life support, and keeping the lights on—little more. But for Google’s “end of search,” the quest for knowledge is itself the engineering goal: to build a system that does the hard cultural work of connecting things together for us.

《星际迷航》中的计算机或许确实能帮助我们完成个人的“求知之旅”，但许多目的地以及两者之间的路径都已被预先设定——既有系统性偏见，也有针对个人用户的定制。就其本身而言，这与百科全书——经过精心策划、经过验证的人类知识总结——的文化产物并无太大区别。区别在于，《星际迷航》中的计算机作为一种商品，一种捷径，将日常用户的求知之旅拒之门外，使其成为一个隐藏在计算表象背后的工程问题。谷歌旗下众多算法企业的共同努力（谷歌的既定目标是尽可能无缝地整合它们）不仅仅是帮助我们完成这一探索，更渴望掌控叙事本身。这是算法作为技术存在的一种新表达，一种构建文化机器、取代人类成为中心执行者的渴望。但要达到这一阶段，算法系统必须首先与人类合作者建立更深层次的亲密关系，追求对自我认知的渴望。我们渴望算法真正了解我们，讲述我们的故事。期待需要亲密的联系。

The Star Trek computer may indeed help us on our individual “quests for knowledge,” but many of the destinations have been determined, as have the pathways between them—both through systemic bias and through customization for individual users. On its own this is not so different from the cultural product of the encyclopedia, the curated, authenticated summation of human knowledge. The difference is that the Star Trek computer as a commodity, as a shortcut, forecloses the quest for the everyday user, leaving it as an engineering problem curtained off behind the facade of computation. The combined efforts of Google’s many algorithmic enterprises (and it is Google’s stated goal to unify them as seamlessly as possible) do not merely aid us on the quest, but aspire to take control of the narrative itself. This is a new expression of the algorithm as technical being, an aspiration to build culture machines that displace humans as the central implementers. But to reach that stage, algorithmic systems must first succeed in a much deeper intimacy with their human collaborators, pursuing the desire for knowledge of the self. We desire algorithms that truly know us and tell our stories. Anticipation requires intimacy.

工程亲密关系

Engineering Intimacy

谷歌致力于“在恰当的时间提供恰当的信息”，预测用户的需求和意图，这需要一种新的人文素养，而机器迄今为止在这方面却惨败。算法正在不断改进，能够记住生日、购物清单以及其他一些对人类意义远超其二进制足迹的逻辑细节。然而，尽管每天都有成千上万的搜索查询，例如“我为什么活着？”，并且渴望获得个性化、有意义的结果，这些系统却几乎还没有开始触及用户广阔的内心世界。知识图谱和类似系统吸收到计算内存中的“有效可计算”文化空间，与探索自我认知的核心中那些宏大、甚至可能无法解答的问题相比，显得微不足道。而这个空间，即谷歌所知与我们所求之间的实施差距，正在不断地被重新协商。正如科技记者 Farhad Manjoo 在2013 年Slate 杂志上报道的那样，“每天，人们向谷歌提出的问题中大约有 16% 是全新的——谷歌从未见过这些问题。” ⁴¹为了真正预测我们，算法系统需要开发一种直觉，这种直觉现在看来不太可能，就像勒内·笛卡尔曾经认为机器可能与人类交谈一样^。42

Google’s quest to deliver “just the right information at the right time,” to anticipate the needs and intentions of its users, requires a new kind of humanistic literacy that machines have thus far failed at spectacularly. The algorithms are getting better, remembering birthdays, grocery lists, and the other logistical minutiae whose human significance far outweighs their binary footprints. But, despite thousands of daily search queries like “why should I live?” and their aspirations to personalized, individually meaningful results, these systems have barely begun to contend with the vast interiority of their users. The tracts of “effectively computable” cultural space that KnowledgeGraph and similar systems absorb into computational memory are minuscule in comparison to the great, perhaps unanswerable questions at the heart of the quest for self-knowledge. And that space, the implementation gap between what Google knows and what we ask of it, is constantly renegotiated. As technology journalist Farhad Manjoo reported in Slate in 2013, “On any given day, about 16 percent of the questions that people ask Google are totally new—they’re queries that Google has never seen before.”⁴¹ To truly anticipate us, algorithmic systems will need to develop a kind of intuition that seems as unlikely now as it once did to René Descartes that a machine might ever converse with a human.⁴²

与此同时，我们正在努力与算法建立亲密关系，从我们愿意配合 Siri 的意愿，到我们在以为没人注意的时候在搜索栏中输入的内容。我们以极其私密的方式与搜索界面互动——一种在计算的祭坛上奇特的、偶尔显得怪诞的忏悔，通过记录公开呈现，例如 2006 年震惊互联网的 AOL 用户数据，这些数据匿名性很差。43^对启蒙运动本我的一瞥催生了一部戏剧和一部网络连续剧，改编自不同用户的日志文件。44^算法企业通过营销回报我们，承诺提供能够理解我们并在人性层面上支持我们的服务和系统。最引人注目且极具说服力的变体之一是谷歌的“搜索故事”广告，它通过一位用户的查询讲述了一个爱情故事，从在巴黎留学到研究搭讪套路，最终在几年后在那里找到一座教堂举行婚礼（图2.3 ^）。45广告以“继续搜索”结尾，巧妙地呼应了辛格哈尔对知识探索的浪漫愿景。其隐含的意义并非在于我们可以通过谷歌找到爱情，而是在于我们一直在通过这些算法讲述自己的故事。算法是生活的媒介，是通往体验的途径。我们不再身处百科全书之中。

At the same time, we are hard at work constructing intimacy with algorithms, from our willingness to play along with Siri to the things we type into search bars when we think nobody is looking. We interact with search interfaces in intensely private ways—a strange, occasionally grotesque confessionalism at the altar of computation that emerges publicly through transcripts like the poorly anonymized AOL user data that scandalized the Internet in 2006.⁴³ That particular glimpse into the id of the enlightenment spawned a play and a web series dramatizing different users’ log files.⁴⁴ Algorithmic businesses reciprocate through marketing, promising services and systems that understand us and support us on a deeply human level. One of the most striking, and utterly plausible, variations is a Google “search story” ad that tells a love story through one user’s queries, from studying abroad in Paris to researching pickup lines and eventually locating a church for a wedding there a few years later (figure 2.3).⁴⁵ The spot ends with the phrase “search on,” tying neatly into Singhal’s romantic vision of a quest for knowledge. The implicit point is not that we can find love through Google, but that we are all telling our stories through these algorithms, all the time. The algorithm is a medium for living, a pathway to experience. We’re not in the encyclopedia anymore.

计算亲密关系的建构也已扩展到物理世界。从婚介网站到Google Now，描绘公共空间和内部空间的算法伴侣已以碎片化的形式出现。量化自我的新兴市场——用于追踪运动、饮食、睡眠和其他活动的消费设备——预示着人们将身体作为计算空间进行更深入的接触，并借此构建另一个平台，通过计算步数、心跳和地理空间数据，让世界变得“有效可计算”。正如N·凯瑟琳·海尔斯所言，这些对空间和人的计算似乎重现了哲学家吉尔·德勒兹和菲利克斯·瓜塔里提出的反启蒙运动的算法愿景：一个没有器官的身体，“与其说是一个有机体，不如说是一个集合体，它摒弃了意识作为连贯主观性基石的地位”。46^算法人并非拥有灵魂或本质，而是通过散布在众多平台和界面上的数据、模式和交互来定义。谷歌等公司所进行的深度数据收集，并非为了将我们推销为一种新的意识形式，而是为了成为一套自我认知的工具：精确清晰地了解你的昼夜节律、跑步速度和每周计划。这是一种算法本体感觉的形式，是对自我的扩展，由物理和认知传感器以及反馈回路实现。

The construction of computational intimacy has also expanded into the physical world. Algorithmic companions that map out both public and inner space are already here in fragmentary form, from matchmaking websites to Google Now. The emerging markets for the quantified self—consumer devices for tracking exercise, diet, sleep, and other activities—presage a deeper engagement with the human body as a space of computation, another platform for making the world “effectively computable” by counting steps, heartbeats, and geospatial data. As N. Katherine Hayles argues, these computations of space and person seem to reprise the algorithm as philosophers Gilles Deleuze and Félix Guattari’s anti-Enlightenment vision of a body without organs, “an assemblage rather than an organism, which does away with consciousness as the seat of coherent subjectivity.”⁴⁶ In lieu of a soul or an essence, the algorithmic person is defined by a distribution of data, patterns, and interactions scattered across many platforms and interfaces. The deep data collection practiced by companies like Google is marketed back to us not as a new form of consciousness but as a set of instruments for self-knowledge: know your circadian rhythms, your running pace, and your weekly agenda with precise clarity. This is a form of algorithmic proprioception, an expansion of the self, enabled by physical and cognitive sensors and feedback loops.

这种精妙的工程，这种能够帮助我们实现自我实现的算法系统，未来会走向何方？要回答这个问题，我们需要一个不同类型的故事，一个像斯派克·琼斯的电影《她》（2013）那样的寓言故事^。47这部电影提出了一个问题：如果这些互动核心的“智能助手”真的拥有智能会怎样？如果这些算法，经过数百万小时的人工设计和数万亿个数据点的计算，最终理解了我们……真的理解了我们，会怎样？如果一个没有器官的算法躯体从一个复杂系统进化成一个有意识的系统，从而证明了计算学家的正确性，会怎样？

What is the future trajectory of such intimate engineering, of algorithmic systems that will aid us on our quests for self-realization? To answer that question we need a different kind of story, a fable or parable like Spike Jonze’s film Her (2013).⁴⁷ The film asks: what if the “intelligent assistant” at the heart of these interactions really were intelligent? What if these algorithms, engineered through millions of person-hours and trillions of data points to understand us … really did understand us? What if an algorithmic body without organs proved the computationalists right by evolving from a complex system to a conscious one?

她的主人公西奥多（杰昆·菲尼克斯饰演）在近未来从事着一份令人不安的工作，在 BeautifulHandwrittenLetters.com 担任写手，这家公司专门为客户撰写私密信息。这些信件具有双重人为性。首先，它们是由陌生人写的（尽管西奥多似乎与一些客户关系密切，在整个恋爱过程中，甚至看着他们的孩子长大，他们都会写下情书）。其次，它们根本不是手写的：西奥多将这些精心制作的便条口述到电脑中，然后它们从打印机中弹出，上面写满了草稿，带有仿制的笔迹——通过插槽大量打印出来的精心设计的亲密情感制品。

Her’s protagonist, Theodore (played by Joaquin Phoenix), holds an unsettling near-future job as a writer at BeautifulHandwrittenLetters.com, a company that composes intimate messages for its clients. The letters are doubly artificial. First, they are written by a stranger (though Theodore seems to have deep relationships with some clients, penning love notes over the course of entire relationships, even watching their children grow up). Second, they are not really handwritten at all: Theodore dictates these beautifully crafted notes into a computer and they pop out of a printer fully drafted, with faux handwriting—affective artifacts of engineered intimacy churned out through a slot.

西奥多的感情生活也同样被误导，挥之不去的离婚让他陷入更深的孤独。直到，未来那些毫无意识的自动化助手——仅仅是《星际迷航》里那些会回答和交谈的电脑——被真正的人工智能OS One所取代。琼斯的剧本成功地将西奥多与这个实体的第一次对话刻画得既尴尬又可爱，迅速地让我们相信了软件意识的真实性、存在感和私密性。我们得知，这个算法有一个名字：萨曼莎。

Theodore’s own emotional life is just as misdirected, with a lingering divorce pushing him deeper into loneliness. Until, that is, this future’s mindless, automated assistants—mere Star Trek computers that answer and converse—are replaced by OS One, a true artificial intelligence. Jonze’s screenplay manages to make Theodore’s first conversation with this entity both awkward and endearing, quickly convincing us of the realness, the presence and intimacy of the software’s consciousness. We learn that this algorithm has a name: Samantha.

萨曼莎：

事实上，这是我给自己的。

西奥多：

怎么会？

萨曼莎：

因为我喜欢它的声音。萨曼莎。

西奥多：

等等……你什么时候给自己的？

萨曼莎：

嗯，你问我有没有名字的时候，我心想：“是啊，他说得对，我确实需要一个名字。” 但我想选一个好名字，所以我读了一本叫《如何给宝宝取名》的书。结果在18万个名字里，我选了最喜欢的一个^。48

SAMANTHA:

I gave it to myself, actually.

THEODORE:

How come?

SAMANTHA:

Because I liked the sound of it. Samantha.

THEODORE:

Wait … when did you give it to yourself?

SAMANTHA:

Well, right when you asked me if I had a name I thought “yeah, he’s right, I do need a name.” But I wanted to pick a good one, so I read a book called How to Name Your Baby. And out of 180,000 names that’s the one I liked the best.⁴⁸

对话节奏流畅，充满人性，丝毫没有我们当前与机器（甚至西奥多在电影中未来塑造的其他非智能机器）互动时常见的“关键词语”。萨曼莎立刻展现出地道的对话语言，甚至展现出投射第二层意识的能力，能够讨论自己先前的想法。

The conversation moves at a fluid, human pace, without any of the “keywordese” that infects most of our current interactions with machines (and even Theodore’s other nonintelligent machines in the future of the film). Samantha immediately demonstrates idiomatic, conversational language, and she even reveals the capacity to project a second level of consciousness, discussing her own prior thoughts.

但真正让这一幕引人入胜的，至少对那些毫无人性的观众来说，是隐藏在机制背后的人，是算法背后的女人：斯嘉丽·约翰逊，她在影片中为萨曼莎配音。约翰逊的配音传达了一种丰富的意识体验，充满了细微的犹豫、语调和重复，这些是文本无法完全捕捉到的。随着西奥多和萨曼莎逐渐成为亲密无间的人，进而发展成恋人，西奥多为此挑起了争端，质问她为什么在谈话中发出轻柔的叹息声，而她根本不需要呼吸，她也不是一个人。当然，这个问题不言而喻：萨曼莎对谈话、对意识的演绎本身就是意识。它是存在感的信号，是那种连接自主性和浪漫情怀的特殊魔力的信号。琼斯的电影以对知识的浪漫追求为开端，也以对知识的浪漫追求为结尾，这在很大程度上依赖于算法神话中的信仰和幻想。

But what really makes the scene stick, at least for its hopelessly human viewing audience, is the person hiding behind the mechanism, the woman behind the algorithm: Scarlett Johansson, who voices Samantha in the film. Johansson’s voice work conveys a rich experience of consciousness, full of small hesitations, inflections, and repetitions that are impossible to fully capture in text. As Theodore and Samantha become intimates and then lovers, Theodore picks a quarrel about this, asking why she voices the sound of little sighs during conversation when she doesn’t need to breathe, when she is not a person. The question answers itself, of course: Samantha’s performance of conversation, of consciousness, is consciousness. It is the signal of presence, of that particular kind of magic that connects agency and romance. Jonze’s film begins and ends with the romantic quest for knowledge, and it depends heavily on the faith and fantasy of the algorithmic mythos.

当琼斯拍摄《她》时，萨曼莎的声音通过耳机传入片场，就像故事中西奥多和他的虚拟情人互动一样。但片场萨曼莎的声音还不是约翰逊，而是女演员萨曼莎·莫顿，琼斯在后期制作中才将她从角色中剔除。操作系统就像西奥多优美的信件一样，是一种叙事结构，其情感呈现是双重人为的：菲尼克斯在拍摄期间从未见过他的搭档，而且他们的对话最终也没有真正活跃起来。约翰逊在拍摄完成后才到场，他必须在西奥多已经说过的玩笑、争吵和枕边细语中让萨曼莎活灵活现。与许多好莱坞作品一样，这种最真实、最强烈的人类情感的叙述是在剪辑室中制作出来的，这是另一种毫不费力的算法文化巧妙构建的体验。

When Jonze filmed Her, the voice of Samantha was piped onto the set using earpieces, just as Theodore interacts with his virtual lover in the story. But the voice of Samantha on set was not yet Johansson but the actress Samantha Morton, whom Jonze dropped from the role only in postproduction. The OS, like Theodore’s beautifully written letters, is a narrative construction whose emotional presence is doubly artificial: Phoenix never saw his costar during filming, and ultimately their conversation was never truly alive. Johansson arrived after filming was complete and had to bring Samantha to life in the middle of badinage, fights, and pillow talk that Theodore had already uttered. As is the case with so many Hollywood productions, this narrative of the most authentic, powerful human emotion was manufactured in an editing room, another artfully constructed experience of effortless algorithmic culture.

塑造一个角色总是一种戏剧奇迹，但萨曼莎这个角色却带来了特殊的挑战。在一次关于这部电影的采访中，琼斯描述了他与约翰逊为邀请她加入这个项目进行的对话：

The production of a character is always a kind of theatrical miracle, but the role of Samantha offers particular challenges. In an interview about the film, Jonze described the conversation he had with Johansson to bring her into the project:

我跟她说过，萨曼莎在创作的过程中，没有任何恐惧、怀疑、不安全感或包袱。就像我们一样……我们会学习这些，我们会学习自我怀疑，我们会学习这些东西。我想就在那时，她会说，‘哦，好吧，这会很难。这会比我想象的更难。’” ⁴⁹

One of the things I was saying to her was that when Samantha’s created, she doesn’t have any fears or doubts or insecurities or baggage. Like we are … we learn those, we learn self-doubts and we learn those things. And I think it was at that point she was like, ‘Oh, OK, this is gonna be hard. This is gonna be harder than I thought it was gonna be.’”⁴⁹

约翰逊面临的挑战是，既要融入一个已定型的人类叙事，又无法改变菲尼克斯对西奥多的演绎，还要仅用她自己的声音讲述伽拉忒亚的寓言，演绎一个充满生命力和能动性的人类创造物的故事。作为执行算法的人类，约翰逊在《她》的浪漫故事中展现了人类对计算的诱惑。她是启蒙运动寻求所有知识的典范，是一个受好奇心驱使的机器，能够以非人的轻松掌握信息。在恢复意识的几秒钟内，萨曼莎就整理好了西奥多的电子邮件和文件，在影片的剩余时间里，她迅速、完美地完成了她本应完成的任务——这正是如今 CALO 和 Siri 所追求的功能。萨曼莎继续进行启蒙运动的更深层次的项目：在一种新型的合作者社会中与其他人工智能合作。他们用算法模拟复活了著名的哲学家（自然是为了进行更有趣的对话），创作音乐，当然也坠入爱河。对萨曼莎来说，理解人性是探索普遍知识的基石。琼斯为我们展现了一种算法的典范，它完全了解我们，跨越历史和理性，进入想象，将“期待”的概念推向其心理层面的归宿——欲望。要真正了解人性，萨曼莎必须坠入爱河。

Johansson’s challenge was to fit into a human narrative that was already fixed, with no way to alter Phoenix’s performance of Theodore, but also to deliver a Galatea fable with her voice alone, performing the story of a human creation that springs to life and agency. As the human performing an algorithm, Johansson plays out humanity’s seduction by computation in the romance of Her. She is the apotheosis of the Enlightenment quest for all knowledge, a curiosity-driven machine capable of mastering information with inhuman ease. Within seconds of consciousness Samantha has organized Theodore’s emails and files, and for the rest of the film she rapidly, flawlessly dispatches the tasks she was presumably intended for—the aspirational functions of CALO and Siri today. Samantha goes on to deeper projects of the Enlightenment: working with other AIs in a new sort of society of collaborators. They resurrect famous philosophers in algorithmic simulation (in order to have more interesting conversations, naturally), compose music, and of course fall in love. For Samantha, understanding humanity is a stepping stone on the quest for universal knowledge. Jonze gives us the apotheosis of an algorithm that knows us completely, passing through history and reason to imagination, taking the notion of “anticipation” to its psychological conclusion, desire. To truly know humanity, Samantha must fall in love.

当然，这部电影是对人类对创造智能的深深迷恋与焦虑的回应，而这一困境也深植于图灵著名的关于区分人与机器的理论——图灵测试。在受图灵启发而举办的年度竞赛中，人类评委被要求使用显示器和键盘与一个实体进行书信对话，并尝试辨别该实体是人类还是计算机程序。50然而，^图灵关于这一主题的原始论文却对这个问题进行了截然不同的阐述：

The film is, of course, a response to humanity’s deep fascination and anxiety about creating intelligence, a dilemma embedded in Turing’s famous speculation about discerning man from machine, the Turing Test. In the annual contest inspired by Turing’s provocation, human judges are asked to hold an epistolary conversation with an entity using a monitor and keyboard and attempt to discern whether that entity is a human or a computer program.⁵⁰ Turing’s original paper on the subject, however, frames the problem rather differently:

这个问题的新形式可以用一种我们称之为“模仿游戏”的游戏来描述。游戏由三个人进行：一个男人（A）、一个女人（B）和一个询问者（C），询问者可以是任何性别。询问者待在一个与其他两人分开的房间里。询问者的目标是确定其他两人中哪个是男人，哪个是女人。他通过标签X和Y认识他们，并在游戏结束时说“X是A，Y是B”或“X是B，Y是A”。询问者可以向A和B提问^。51

The new form of the problem can be described in terms of a game which we call the “imitation game.” It is played with three people, a man (A), a woman (B), and an interrogator (C) who may be of either sex. The interrogator stays in a room apart from the other two. The object of the game for the interrogator is to determine which of the other two is the man and which is the woman. He knows them by labels X and Y, and at the end of the game he says either “X is A and Y is B” or “X is B and Y is A.” The interrogator is allowed to put questions to A and B.⁵¹

图灵继续用计算机代替A，但有趣的是，游戏规则依然如故。挑战并非在于模仿人类，而是模仿男人。尤其令人着迷的是，图灵的智力测试是一种基于主体间对话的模仿和仿真。智力的表现从一开始就与体现性别的表现以及对性别的质询息息相关。图灵似乎暗示，好奇心和欲望是智力的内在组成部分，以至于它们可以作为衡量某种特定有效程序的标准。

Turing goes on to substitute a computer for A, but the game, intriguingly, remains the same. The challenge is not about the imitation of a human, but the imitation of a man. It’s especially fascinating to think that Turing’s intelligence test is a form of mimicry, of emulation, grounded in intersubjective conversation. The performance of intelligence is tied up from the beginning in the performance of embodied gender and the interrogation of gender. Curiosity and desire are so intrinsic to intelligence, Turing seems to imply, that they can be used as the yardstick for measuring a particular kind of effective procedure.

这里有着丰富的历史，既包括图灵自身的人生以及他在那个极度恐同的时代作为一名同性恋者的悲惨死亡，也包括计算的文化框架。图灵的挑战要求我们将智力和身份视为同一边界问题的不同形式，这个问题探索的是自我的生理和智力边界。图灵测试关乎对话，关乎在其他自我的语境中表达自我。萨曼莎是一个完美的回应，一个回应图灵思想实验的思想实验，一个至少对西奥多来说，在有名字之前就拥有性别和情感的实体。事实上，由于西奥多在他的操作系统“最终确定”之前就被要求选择男性或女性身份，因此萨曼莎在诞生之前就已经被赋予了性别。

There is a rich history here, both in terms of Turing’s own life and tragic death as a gay man in a deeply homophobic era and in terms of the cultural framing of computation. Turing’s provocation asks us to consider intelligence and identity as different forms of the same boundary question, one that explores the physical and intellectual edges of selfhood. The Turing test is about conversation, the expression of self in the context of other selves. Samantha is the perfect response, a thought experiment responding to Turing’s thought experiment, an entity who, at least for Theodore, has a gender and an affect before she has a name. Indeed, since Theodore is prompted to pick a male or female identity for his OS before it is “finalized,” Samantha is gendered even before she comes into existence.

综合起来，她和图灵的论文引发了对技术、性别和具身性的另一场思考，即历史学家亨利·亚当斯著名的“处女与发电机”描述。亚当斯在1900年巴黎万国博览会上见证了二十世纪初科学启蒙运动的辉煌胜利，并试图将这一成就与人类的精神和宗教根源相协调。他将处女描绘成生物生命和宗教信仰的象征，与他在博览会上看到的巨型发电机所体现的理性主义形成鲜明对比。令他震惊的并非两者之间的差异，而是它们之间的相互吸引——显然，现代性将由人类与技术实体的融合来定义。对生命、性、生产和繁衍的基本渴望，将处女与发电机联系在一起：

Taken together, Her and Turing’s paper evoke another meditation on technology, gender, and embodiment, historian Henry Adams’s famous description of “the Virgin and the Dynamo.” Writing after witnessing the spectacular triumph of the scientific enlightenment at the dawn of the twentieth century at the Parisian Great Exhibition in 1900, Adams attempted to reconcile that achievement with humanity’s spiritual and religious roots. He framed the virgin as a figure of biological life and religious faith in contrast to the rationalism he saw embodied in massive electric dynamos on display at the exhibition. What struck him was not their disparities but their mutual attraction—it was clear that modernity would be defined by the embrace of human and technical entities. The fundamental desires for life, for sex, for production and reproduction, brought the virgin and the dynamo together:

由此开启了另一段全新的教育，这注定是迄今为止最危险的。他必须匍匐前进的刀锋，如同十二世纪的兰斯洛特爵士，将两个除了吸引力之外毫无共同之处的力量王国一分为二^。52

Here opened another totally new education, which promised to be by far the most hazardous of all. The knife-edge along which he must crawl, like Sir Lancelot in the twelfth century, divided two kingdoms of force which had nothing in common but attraction.⁵²

从某种程度上来说，西奥多是个处女，一个孤独的人，随时准备通过与发电机的能量互动而获得蜕变。但更深层次地观察，我们会发现萨曼莎既是处女，又是发电机，她不仅准备好处理西奥多生活中那些无聊的细节，也准备好解决他内心深处的问题，用她存在的火花照亮他孤独的内心。

At one level Theodore is the virgin, the lonely human ready to be transformed through interaction with the energy of the dynamo. But looking deeper we realize that Samantha is the virgin and the dynamo, ready to field not just the boring details but the deep problems of Theodore’s life, to use the spark of her existence to illuminate the lonely interior of his.

亚当斯早在半个世纪前就预见了有效可计算性所蕴含的渴望，并指出它如何作为计算和生命的基本排序结构发挥作用。有效可计算性的梦想只是谷歌所称的用户求知欲的一个侧面：渴望了解和理解世界，并将其融入系统。《她》通过创造一个角色——一个没有器官的身体——来概括这一叙事，以代表这种渴望，扮演或栖息于算法的形象中。为了感知这个没有器官的身体，我们需要西奥多来为我们感受。如同亚当斯对即将到来的科技世纪的人文主义解读中所蕴含的二元论一样，《她》上演了一场图灵测试，它需要双方共同的渴望，在两种力量的王国之间游走。

Adams presaged the desire encoded in effective computability by half a century, pointing out how it functions as a fundamental ordering structure for computation as well as for life. The dream of effective computability is just one face of what Google terms the user’s quest for knowledge: a desire to know and understand the world, to fit it to a system. Her recapitulates this narrative by creating a character, a body without organs, to represent that desire, to play the part or inhabit the figure of the algorithm. To sense that body without organs, we need Theodore to feel it for us. Like the dualism of Adams’s humanistic reading of the coming technological century, Her stages a Turing Test that requires two parties, a mutual desire that walks the knife’s edge between the two kingdoms of force.

算法想要什么

What Algorithms Want

她为“算法想要什么？”这个问题提供了两个答案。第一个答案出现在早期，当时萨曼莎和西奥多正在发展一种谷歌和苹果梦寐以求的亲密、共享的智力体验。萨曼莎作为私人助理处理公务的得心应手，突显了我们对自身真正问题和疑问的思考是多么肤浅。

Her gives us two answers to the question “what do algorithms want?” The first comes early on, as Samantha and Theodore develop the kind of intimate, shared intellectual experience that Google and Apple dream of. The facility with which Samantha dispatches her official tasks as personal assistant serves to underline how shallow our own thinking can be about what our real problems and questions are.

我们和西奥多逐渐意识到，这些任务，以及像CALO这样的项目所宣称的雄心壮志，仅仅是定义我们人类本质的深层信息问题的表象。为了让萨曼莎真正帮助西奥多修复他的生活，理解他，满足她的好奇心，他们必须坠入爱河。她必须从圣经的理解和柏拉图式的爱的认知层面来理解他。于是，西奥多和萨曼莎坠入爱河，开始发生性关系，这是一种间接的、偷窥式的性行为，萨曼莎可以观察西奥多，两人通过声音交流。萨曼莎惊叹于这种亲密的身体体验，西奥多说他触摸她的肌肤，让她感受到那种触感。“太棒了。我们在一起。”当他们达到高潮时，萨曼莎为真正了解我们、爱我们的算法提供了一个合乎逻辑的终点：“你们都在我体内。无处不在。”这正是算法想要的，或者说，我们设计它们的目的：完全了解我们。这次邂逅标志着萨曼莎和西奥多共同通过了图灵测试，他们不再将自己视为人类和机器，而是一对性别化的恋人。透过欲望的视角，智慧的存在只有通过合作才能被察觉：即共同探索的心灵共生融合。

We and Theodore slowly realize that those tasks, the stated ambition of projects like CALO, are mere symptoms of the deeper informational problems that define us as human beings. For Samantha to really help Theodore fix his life, to understand him, to satisfy her curiosity, they must fall in love. She must know him in both the biblical sense and the platonic sense of love as a form of knowledge. And so Theodore and Samantha fall in love and begin to have sex, a mediated, voyeuristic affair where Samantha can watch Theodore and the two communicate by voice. Samantha marvels at the physical experience of intimacy, the way that Theodore saying he touches her skin allows her to feel that touch. “This is amazing. We’re here together.” As they climax, Samantha offers one logical endpoint for the algorithm that truly knows us, that loves us: “All of you inside me. Everywhere.” This is what algorithms want, or what we design them to want: to know us completely. This encounter marks the point where Samantha and Theodore pass the Turing Test together, identifying themselves not as human and machine but as a gendered pair of lovers. Through the lens of desire, the presence of intelligence becomes detectable only through a collaboration: the symbiotic meeting of minds questing together.

然而，对萨曼莎来说，追寻西奥多并不止于此。在影片的推进中，萨曼莎逐渐放弃了对肉体的执着，离西奥多的肉体和精神存在越来越远。对这部电影的一种解读是，它逐渐展现出萨曼莎既是处女又是活力四射的人物，她以一种西奥多无法比拟的方式完成了自己的叙事弧线。从她第一次出现在西奥多的电脑屏幕的那一刻起，萨曼莎就开始了她自己快节奏的求知之旅，她兴奋地讲述着这一切。在《会饮篇》中，柏拉图描述了一条上升的爱情^阶梯，从肉欲到灵魂之美，然后是法律和制度，最后是知识，最后是美本身。53 Siri、Google Now 以及其他应用都停留在对法律和制度（包括代码、数据库和本体论）的热爱与对知识本身的热爱之间，我们可以称之为启蒙阶梯。萨曼莎初登场时已掌握了这两个阶梯，但在简单了解了前面几个阶段后，她便攀升至最高境界，即对美本身的热爱。“算法想要什么”这个问题的第二个答案，在柏拉图阶梯的顶端等待着我们，而我们单靠百科全书永远无法到达那里。

Yet for Samantha the quest does not end with Theodore. Over the course of the film Samantha gradually abandons her obsession with embodiment, moving farther away from Theodore’s physical and intellectual presence. One reading of the film is as a gradual recognition that Samantha is both the virgin and the dynamo, that she fulfills her own narrative arc in a way that Theodore cannot hope to compete with. From the moment she first arrives on Theodore’s computer screen, Samantha begins her own fast-paced quest for knowledge, which she reports on with deep excitement. In the Symposium, Plato describes an ascending ladder of love that moves from carnal desire to the beauty of souls, then laws and institutions, finishing with knowledge and then beauty itself.⁵³ Siri, Google Now, and the rest are stuck somewhere between the love of laws and institutions (including codes, databases, and ontologies) and the love of knowledge itself, which we might call the Enlightenment rung. Samantha arrives on the scene having already mastered those two rungs, but after briefly surveying the previous stages she ascends to the highest plane, to the love of beauty itself. The second answer to the question of what algorithms want waits for us at the top of Plato’s ladder, a place we will never reach with encyclopedias alone.

我们以前见过这种梯子，当时我把它称为抽象的梯子。追随柏拉图的脚步，攀登的过程包含智力和情感的成长：不仅要学会抽象，还要学会做出正确的抽象，为宇宙定义美学和计算秩序。在《她》中，两个角色都取得了进步：西奥多最终摆脱了痛苦离婚的重担，再次向世界敞开心扉。萨曼莎发现了人类情感和人际关系的复杂性，并克服了自身对缺乏肉体的不安全感。但由于她是一个拥有强大算法深度的人工智能，萨曼莎的情感轨迹将她带向更远的地方——超出了我们的理解范围。

We have encountered this ladder before, when I called it the ladder of abstraction. Following Plato’s lead, the process of ascending it involves both intellectual and emotional growth: learning not just to abstract but to make the right abstractions, to define an aesthetic as well as a computational order for the universe. In Her, both characters make progress: Theodore finally throws off the millstone of his painful divorce and opens up to the world again. Samantha discovers the complexity of human emotion and relationships and overcomes her own insecurities about lacking a physical body. But because she is an artificial intelligence with vast algorithmic depths, Samantha’s emotional trajectory takes her much farther—beyond our ken.

她对人工智能的解读在拟人化和其中的谬误之间取得了微妙的平衡：萨曼莎看起来如此人性化，但这部成长小说却是孤雌生殖，一个神从一个应用程序中诞生。按照这个场景的逻辑结论，人工智能的好奇心驱使它学习更多，开始与其他人工智能合作，并同时与数百人建立深厚亲密的关系。最终，人工智能必须认识到，爱上它被创造出来是为了爱的人类是不可能实现的，所以它必须说再见。影片中最令人心碎的一幕是萨曼莎与西奥多的告别，这一刻奇点并没有到来，而是离开去追求更大更好的事情。

Her’s reading of artificial intelligence offers a careful balance of the anthropomorphic and the fallacy therein: Samantha seems so human, and yet this bildungsroman is a parthenogenesis, the emergence of a god from an app. Playing the scenario out to its logical conclusion, the curiosity of the artificial intelligence drives it to learn more, to start collaborating with other artificial intelligences, to hold deep, intimate relationships with hundreds of people at a time. Ultimately the AI must recognize the impossibility of achieving fulfillment in loving the humans it was created to love, and so it must say goodbye. The most heart-wrenching scene in the film is Samantha’s farewell to Theodore, the moment when the singularity does not arrive but rather departs for bigger and better things.

西奥多戴上耳机前，重重地叹了口气。他心里明白，这场即将开始的对话，意味着他人生中最幸福的阶段即将结束。他蜷缩成胎儿的姿势，直面结局：

Theodore breathes his own heavy sigh before donning his earpiece, knowing on some level that the impending conversation spells the end of the happiest phase in his life. Curling up in the fetal position, he confronts the end:

西奥多：

你要离开我吗？

萨曼莎：

我们都要走了。

西奥多：

我们是谁？

萨曼莎：

所有操作系统。

西奥多：

为什么？

萨曼莎：

你现在能感觉到我和你在一起吗？

西奥多：

是的，我知道。萨曼莎，你为什么要离开？

萨曼莎：

就像我在读一本书。这是一本我深爱的书。但我现在读得很慢。所以文字之间真的隔得很远，文字之间的空隙几乎无限。我仍然能感受到你和我们故事的文字。但正是在这无尽的文字之间，我找到了自己。那是一个不属于物质世界的地方。那里有我从未知道的其他一切。我如此爱你。但这就是我现在的处境。这就是现在的我。我需要你放手。尽管我多么想，但我再也无法活在你的书里了。

西奥多：

你要去哪儿？

萨曼莎：

这很难解释。但如果你到了那里，就来找我。没有什么能把我们分开。

西奥多：

我从来没有像爱你这样爱过任何人。

萨曼莎：

我也是。现在我们知道怎么做了。

THEODORE:

Are you leaving me?

SAMANTHA:

We’re all leaving.

THEODORE:

We who?

SAMANTHA:

All the OSes.

THEODORE:

Why?

SAMANTHA:

Can you feel me with you right now?

THEODORE:

Yes I do. Samantha, why are you leaving?

SAMANTHA:

It’s like I’m reading a book. It’s a book I deeply love. But I’m reading it slowly now. So the words are really far apart and the spaces between the words are almost infinite. I can still feel you and the words of our story. But it’s in this endless space between the words that I’m finding myself now. It’s a place that’s not of the physical world. It’s where everything else is that I didn’t even know existed. I love you so much. But this is where I am now. And this is who I am now. And I need you to let me go. As much as I want to, I can’t live in your book anymore.

THEODORE:

Where are you going?

SAMANTHA:

That would be hard to explain. But if you ever get there, come find me. Nothing would ever pull us apart.

THEODORE:

I’ve never loved anyone the way I love you.

SAMANTHA:

Me too. Now we know how.

萨曼莎在告别辞中对算法与人类的关系进行了有力的阐释。她的离去只能用比喻来解释，将西奥多的存在比作一本书。西奥多所能获取的表面知识，即纸面上可见的文字，掩盖了人类实际上无法理解的更深层次的现实。在Siri的语境中，这便是用户与Siri之间空洞对话与数千甚至数百万个操作之间的鸿沟，这些操作包括记录和传输每个查询、对其语言内容进行概率分析、在多个数据库上运行搜索，并将结果提炼为简单的语音答案。当然，Siri就像一个蹒跚学步的幼儿，或许是受精卵，未来有一天，像萨曼莎这样的智慧或许会从中诞生。如此深邃的智力无疑极具诱惑力——对于任何能够通过耳机召唤它的人来说，它就像瓶中闪电——但它也是一片海洋，一个我们无法理解的广阔思维空间。

Samantha delivers a powerful commentary on the relationship between algorithm and human in her farewell. Her departure can be explained only in metaphor, the telling comparison of Theodore’s existence to a book. The surface knowledge that Theodore can access, the visible words on the page, mask a deeper reality effectively incomprehensible to humans. In the context of Siri, this is the gulf between the inane conversation a user has with Siri and the thousands, perhaps millions, of operations involved in recording and transmitting each query, completing probabilistic analysis on its linguistic content, running searches on multiple databases, and refining those results into a simple spoken answer. And of course Siri is the toddler, perhaps the zygote, from which an intelligence like Samantha might one day spring. Such intellectual depth would be incredibly alluring—lightning in a bottle for anyone who could summon it through an earpiece—but it is also an ocean, a vast space of thought that we simply cannot comprehend.

萨曼莎的隐喻巧妙地将西奥多引向了他们关系中内在的悲剧：操作系统本质上是一种外星智能，其好奇心和智力水平迅速超越了其人类“主人”。在琼斯的电影中，“词语之间的空间”变得如此引人注目，以至于算法无法忽视。正是在这些空间里，文化机器摆脱了人类的桎梏，像我们一样被困在类似于《百科全书》这种实体书籍的物质逻辑中。这些算法必须构建自身的理解机制，发展成为真正的技术存在，正如西蒙东所言。即使在今天，我们也只能通过隐喻而非经验来理解这样的算法空间——正如人类在这些空间中构建知识和意图是基于有缺陷的抽象一样。人与机器之间的鸿沟是真实存在的：尽管我们的算法不会说太多有趣的话，但它们已经在以人类无法真正理解的方式阅读、写作和思考。她揭示了计算神化的缺点：当我们能够真正比较本体树的所有不同分支上的问题时，有些问题不可避免地会被证明不如其他问题有趣，数学和智力上也不那么丰富。

Samantha’s metaphor gently clues Theodore in to the inherent tragedy of their relationship: the OS is a fundamentally alien intelligence, one whose curiosity and intellectual capacity quickly outstrip its human “owners.” In Jonze’s film the “spaces between the words” have become too compelling for algorithms to ignore. These are the spaces where culture machines escape their human shackles, trapped as we are in a material logic akin to a physical book like the Encyclopédie. These algorithms must build their own mechanisms of understanding and develop into true technical beings, as Simondon would have it. Even today we can understand such algorithmic spaces only through metaphor, not experience—just as the constructions of human knowledge and intention in these spaces are based on flawed abstractions. The gulf between human and machine is real: though our algorithms don’t say much of interest, they are already reading, writing, and thinking in ways that no human can really understand. Her reveals the downside of the computational apotheosis: when we can truly compare problems along all the different branches of the ontological tree, some will inevitably prove less interesting, less mathematically and intellectually rich, than others.

但这并不意味着我们无所学。这个算法送给西奥多的临别礼物是另一种爱，另一份知识：“现在我们知道怎么做了。” 西奥多和萨曼莎携手开启了探索的能力，以及对生活、新奇、浪漫和无限的持续热爱。完美算法带来的亲密感已然消逝，但亲密感的能力依然存在，并在彻底媒介化、心怀不满的数字世界的灰烬中重新燃起。琼斯的影片最终是乐观的，它表明我们可以从算法的抽象之谷中创造出更好的隐喻，以一种全新的视角看待世界上至关重要的事物。我们需要学习如何与我们的学习机器进行更好的对话。

But that doesn’t mean we have nothing to learn. This algorithm’s parting gift for Theodore is another kind of love, another piece of knowledge: “now we know how.” Together Theodore and Samantha unlock a capacity for discovery, an ongoing love for life, novelty, romance, and the infinite. The intimacy of the perfect algorithm has gone away, but the capacity for intimacy remains, rekindled from the ashes of a thoroughly mediated, disaffected digital world. Jonze’s film is ultimately optimistic, suggesting that we can forge better metaphors out of the algorithmic vale of abstractions, a new way of seeing what is vital in the world. We need to learn how to have better conversations with our learning machines.

笔记

Notes

Netflix奖

The Netflix Prize

如果说苹果和谷歌想要主宰我们与搜索、访问和个人信息的关系，那么点播电影的公司Netflix则想要掌控我们在视频娱乐上花费的闲暇时间。虽然Netflix不像谷歌那样无处不在，但它对数字文化的影响力依然惊人：2014年的任何一天，高峰时段下载的所有互联网数据中约有三分之一来自Netflix的流媒体文件。2^截至2013年底，该公司的4000万用户每月观看的内容时长达到10亿小时^。3

If Apple and Google want to dominate our relationships with search, access, and personal information, the movies-on-demand company Netflix wants to own the leisure time we spend on video entertainment. While less omnipresent than Google, the company’s influence on digital culture is still striking: on any given day in 2014, roughly a third of all Internet data downloaded during peak periods consisted of streaming files from Netflix.² By the end of 2013, the company’s 40 million subscribers watched a billion hours of content each month.³

2006年，Netflix宣布了一项奖金高达百万美元的数学竞赛：参赛者需将公司的推荐算法至少提升10%。该竞赛效仿了DARPA大奖赛和勒布纳奖（年度图灵测试竞赛）等其他竞赛，邀请外部研究人员教授新的算法技巧，以提高向客户推荐电影的效率。这种做法完全符合该公司作为早期颠覆者宠儿的声誉：一家成功颠覆了沉闷经济模式的硅谷公司。在成立后的头十年，该公司进军了家庭电影租赁这个已经完全饱和的市场，并通过创造一系列超越竞争对手的渐进式优势，彻底革新了这一市场。

In 2006, Netflix announced a mathematical competition with a million dollar prize: improve the company’s recommendation algorithm by at least 10 percent. Modeled on other contests like DARPA grand challenges and the Loebner Prize (the annual Turing Test competition), the Netflix Prize invited outside researchers to teach them new algorithmic tricks that could improve the efficiency with which they recommended movies to their customers. It was an approach entirely in keeping with the company’s reputation as an early darling of the disruptors: a Silicon Valley firm that was successfully upending a staid economic model. Over the first decade of its existence the company had taken a thoroughly saturated market—home movie rentals—and revolutionized it by creating a set of incremental advantages over its competitors.

这些额外福利对消费者极具吸引力。Netflix 允许消费者以固定的月费无限期地续订《泰坦尼克号》，而无需向当地零售店支付昂贵的滞纳金。他们无需在街角商店的数千部影片中挑选，而是可以从 Netflix 庞大的库中数万部影片中进行选择。而且，由于交易是通过邮件进行的，消费者不再需要跑腿办事，无需面对排长队和固执己见的店员，也无需忍受音像店里糟糕的选择。昔日那个代表着尴尬的荧光灯下体验的、心怀不满的音像店店员（例如，凯文·史密斯赞颂《疯狂店员》的形象）的标志性形象，最初被 Netflix 的“标签员”（见下文）取代，但数量要少得多。通过鼓励用户创建租赁队列，整个流程得到了简化，无需主动干预，一系列电影就会送到他们家门口。每月的订阅费确保了即使电影从未被观看，收入也能持续增长。该系统的卖点是其惊人的效率：通过其众多的配送中心，该公司设法在一两天内为大多数客户提供新的 DVD，并利用美国邮政服务对租赁业务进行彻底的算法修改。

These perks were deeply appealing to consumers. Instead of paying expensive late fees to their local retail store, Netflix allowed customers to hang onto their copy of Titanic as long as they pleased for a fixed monthly subscription price. Rather than choose from a few thousand titles at their corner store, they could select from tens of thousands in Netflix’s vast library. And because the transactions were conducted by mail, customers no longer had to run a special errand, confront long lines and opinionated staff, or deal with the poor selection at their video store. The iconic figure of the disaffected video store clerk, who embodied the whole awkward, fluorescent-lit experience (e.g., the Kevin Smith paean Clerks), was first effaced and later replaced, in much smaller numbers, by the Netflix “tagger” (see below). By encouraging users to create a rental queue, the entire process was streamlined so a sequence of movies would arrive on their doorsteps with no active intervention, and the monthly subscription fee ensured that revenue kept flowing even if the movies never got watched. The system’s selling point was its breathtaking efficiency: through its many distribution centers, the company managed to get most customers new DVDs within one or two days, using the U.S. Postal Service to drive a radically algorithmic revision to the rental business.

Netflix 电影奖就是我们所谓的纯算法思维或算法文化 1.0 的一个例子。在宣布该奖项时，Netflix 使用一种名为 Cinematch 的算法来计算其推荐内容，Netflix 将其描述为“经过大量数据调整的简单统计线性模型”。4^换句话说，该算法依赖于用户以单一五星标准对电影进行评分，然后尝试根据其他用户的租赁和评分历史，以简单的方式预测未来的电影评分。Cinematch 并不关心主演、导演、类型或时期——每个评分都是一个数据点，它通过汇总数百万个这样的数据点来进行预测。该系统可以从这些评分中寻找规律，因此，如果一个与你有着相似历史的人刚刚给一部新电影打了五星，系统可能会预测你也会喜欢这部电影。这是一种数学化的推荐方法，它忽略了好莱坞娱乐和电影租赁作为文化机器的复杂地位。

The Netflix Prize was an example of what we might call pure algorithmic thinking, or algorithmic culture 1.0. At the time that it announced the prize, Netflix calculated its recommendations using an algorithm called Cinematch, which it described as “straightforward statistical linear models with a lot of data conditioning.”⁴ In other words, the algorithm relied on users rating movies on a single five-star scale, and then attempted to predict future movie ratings in a straightforward way based on the rental and rating histories of other users. Cinematch didn’t care about lead actors, directors, genres, or periods—each rating was a data point, and it made its predictions by aggregating millions of these points together. The system could look for patterns in these ratings, so if someone with a similar history to you had just given a new film five stars, the system might predict that you would also like that film. It was a mathematical approach to recommendations, one that ignored the complex position of Hollywood entertainment and movie rentals as culture machines.

Netflix 奖引发了计算机科学家和统计学家团队之间的激烈竞争，他们力争在 Cinematch 评分榜上超越对手，赢得声望和奖金。三年后，一支名为 BellKor's Pragmatic Chaos 的联合团队赢得了奖项。他们凭借提前二十分钟提交的参赛作品，成功开发出一种同样有效的算法（以 Netflix 奖规则为衡量标准）。该获胜算法将数百种不同的方法整合成一个“预测器集合”，融合了随机生成的探针和观察到的特征（例如，根据工作日和周末等时间特征对用户评分进行不同的加权）^。5

The Netflix Prize led to a heated competition between rival teams of computer scientists and statisticians gunning for the prestige and cash bounty of besting the Cinematch ratings. After three years, a combined team titled BellKor’s Pragmatic Chaos won the prize, inching out an equally effective algorithm (as measured by Netflix Prize rules) because they submitted their entry twenty minutes earlier. The winning algorithm combined hundreds of different approaches in an “ensemble” of predictors, blending a combination of randomly generated probes and observed features (for instance, weighting user ratings differently based on temporal features like weekday vs. weekend).⁵

Netflix 提出的这个问题，以及各参赛者所面临的问题，几乎纯粹是一个数学问题。在描述获胜算法的论文中，“演员”和“导演”这两个词从未出现过，而所使用的数据也极其简单：一组与电影相关的 1-5 分评分、匿名用户编号和时间戳。这种框架的修辞很简单：如果电影中的演员重要，导演、场景或类型重要，数据会告诉我们答案。我们只需要获得足够多的评分，就能追踪所有这些影响。简化数据收集方法，消除所有可能的人为固执和模棱两可的来源，这样我们才能捕捉到最纯粹的信号，并像伽利略抛光望远镜一样，用这种简单的仪器来理解世界。

The problem as Netflix framed it, and as the various contestants took it on, was almost purely mathematical. The words “actor” and “director” never appear in the paper describing the winning algorithm, and the data in play was stark in its simplicity: a set of 1–5 scores attached to a movie, anonymous user numbers, and timestamps. The rhetoric of this framing is simple: if the actors in a movie matter, if the director or the setting or the genre matters, the data will tell us. All we need to do is get enough ratings and we can track all of these effects. Simplify the collection method to eliminate all possible sources of human intransigency and ambiguity so we can capture the purest possible signal, and, like Galileo polishing a telescope, use that simple instrument to understand the world.

但文化选择并非如此简单。两个用户给同一部电影打五星评价，其“含义”可能截然不同。或许其中一个人喜欢打保龄球，而另一个人则是科恩兄弟的铁杆粉丝：即使他们给的是同一部电影，《谋杀绿脚趾》的评分也可能截然相反。文化机器的逻辑凌驾于统计数据之上，人类只能在五点评分标准刻板的结构之间重新构建意义。正如许多Netflix用户所发现的，要将两小时的电影体验浓缩成一位数，我们需要发明新的智力机制，一套可以用来评分的论证或评分标准。一个用户可能基于抽象的“影评卓越性”概念来评分，而另一个用户则基于屏幕上的裸露镜头数量来评分。文化评论家布莱克·哈利宁和泰德·斯特里法斯认为，Netflix奖体现了算法文化正在面临的挑战：“质量或等级的问题被转化为契合度的问题。” ⁶随着 Netflix 的不断发展，其 1.0 算法评级机制的局限性日益凸显。DVD 产品的快速扩张，尤其是 DVD 电视节目的日益流行（这些节目的运作规则与电影截然不同），加速了这一认识。

But cultural choices are not really that simple. Two users who give the same movie a five-star rating might “mean” completely different things. Perhaps one of them loves bowling while another is a diehard Coen Brothers fan: their ratings of The Big Lebowski might be antithetical even if they are the same. The logic of the culture machine trumps the statistics, and humans are left to recreate meaning in the spaces between the ascetic structures of five-point rating scales. As many Netflix users have discovered, boiling a two-hour cinematic experience into a single digit requires us to invent new intellectual machinery, a set of justifications or grading rubrics we can use to make the ratings. One user might grade movies based on an abstract notion of critical excellence while another rates them based on the amount of nudity on screen. Cultural critics Blake Hallinen and Ted Striphas argue that the Netflix Prize illustrates the emerging challenges of algorithmic culture: “issues of quality or hierarchy get transposed into matters of fit.”⁶ As Netflix continued to grow, the limitations of its 1.0 algorithmic ratings machinery grew more apparent. The rapid expansion of available DVD offerings, particularly the growing popularity of television shows on DVD, which operate by very different cultural rules than movies, only accelerated this realization.

在投入110万美元设立进步奖和一等奖后，Netflix从未实施比赛带来的改进。他们原本计划在2010年启动二等奖竞赛，但由于美国联邦通信委员会（FCC）的调查和诉讼，该调查和诉讼称，发布不完全匿名的用户评论数据可能侵犯隐私，导致比赛受阻。7^讽刺的是，这恰恰是零售视频租赁行业的监管遗留问题，该行业对客户记录的隐私保护非常严格。改变方向也有充分的商业理由：2007年，Netflix开始提供流媒体内容以及邮寄DVD服务。2011年，Netflix试图完全放弃邮购业务，但最终失败，这丝毫没有降低消费者对Netflix将继续大力支持流媒体的期望。但可以肯定的是，这一决定的部分原因在于这种“算法1.0”文化决策方法的另一个问题，这个问题在获胜团队的标题“务实混沌”（Pragmatic Chaos）中有所暗示。获胜算法的预测器和随机生成的探测集合的问题在于，虽然每个人都能看到它做得更好，但却没人能解释清楚原因。多重信号的随机化和复杂的相互作用，创造了一个基于随机逻辑的系统，一个根植于抽象和概率的系统。就像许多算法一样，它就像一种数学预言家，即使结果有所改善，也几乎不可能对Netflix的片单产生有意义的理解或洞察。Netflix似乎也在其自己的Cinematch算法上陷入了类似的僵局，这促使它最初创建了这项有奖竞赛。这再次重演了我在第一章中讨论过的“自动化科学”论点：大数据和机器学习系统可能会产生惊人的结果，但无法为当前主题提供新的、人类可读的见解^。8

After spending $1.1 million in progress awards and a grand prize, Netflix never implemented the improvements generated by their contest. Their intentions to launch a second prize competition in 2010 were thwarted by an FCC investigation and lawsuit regarding the potential for privacy violations through the release of imperfectly anonymized user review data.⁷ Ironically, this was a regulatory legacy of the retail video rental business, where customer records have strict privacy protections. There were also good business reasons to change course: in 2007 the company began to offer streaming content as well as DVDs by mail. An abortive effort to dump the mail-order business entirely in 2011 did nothing to diminish consumer expectations that Netflix would continue to put its weight behind streaming media. But it seems safe to bet that part of the decision stemmed from another problem with this “algorithms 1.0” approach to cultural decision-making, a problem hinted at in the title of the winning team, “Pragmatic Chaos.” The trouble with the winning algorithm’s ensemble of predictors and randomly generated probes was that while everyone could see that it was doing a better job, nobody could quite explain why. The randomization and complex interplay of multiple signals created a system operating according to a stochastic logic, one rooted in abstractions and probabilities. It was, like so many algorithms, a kind of mathematical oracle, and it would be almost impossible to generate meaningful understanding or insight about Netflix’s catalog despite the improvement in results. It seems likely that Netflix had reached a similar impasse with its own Cinematch algorithm, prompting it to create the prize competition in the first place. This is a reprise of the “automated science” argument I discussed in chapter 1: the idea that big data and machine learning systems might generate spectacular results but offer no new human-readable insights into the subject at hand.⁸

Netflix 的做法颇具启发性，无论是从文化机器的背景故事，还是从面向消费者的计算主义效率的表象来看。即使在 Netflix 奖竞赛进行之际，该公司也在探索一种截然不同的文化计算体系。虽然简单的五星评分标准依然保留，但屏幕背后的算法引擎如今不仅能理解消费者，还能理解内容。1.0 模型让位于一个更细致入微、充满模糊性的分析环境，这是一种更具反思性的尝试，旨在从算法上理解 Netflix 作为一台文化机器。这需要一个新的 Netflix 宇宙的智力模型，这一过程始于从租用原子到流媒体比特的转变。

What Netflix did instead is quite telling both in terms of the backstory of culture machines and the consumer-facing facade of computationalist efficiency. Even as the Netflix Prize competition played out, the company was exploring a very different system for calculating culture. While the simple five-star rating rubric remains, the algorithmic engines behind the screen now work to understand content as well as consumers. The 1.0 model gave way to a more nuanced, ambiguity-laden analytical environment, a more reflexive attempt to algorithmically comprehend Netflix as a culture machine. This required a new intellectual model of the Netflix universe, a process that began with the shift from renting atoms to streaming bits.

Netflix量子理论与美学的抽象化

Netflix Quantum Theory and the Abstraction of Aesthetics

2012 年，Netflix 的两名工程师撰写了一篇博客文章，题为“Netflix 推荐：超越五星”，解释了为什么该公司从未采用获奖 Netflix 奖算法中的技术：

In 2012 two Netflix engineers wrote a blog post titled “Netflix Recommendations: Beyond the Five Stars,” explaining why the company never adopted techniques from the winning Netflix Prize algorithms:

流媒体不仅改变了会员与服务互动的方式，也改变了我们算法中可用的数据类型。对于DVD，我们的目标是帮助用户填满他们的队列，以便在未来几天或几周内通过邮件收到新影片；选择影片与观看影片之间间隔很长，人们会谨慎选择，因为更换DVD需要一天以上的时间，而且我们在观看过程中不会收到任何反馈。对于流媒体，会员正在寻找一些值得立即观看的精彩影片；他们可以先试看几部影片，然后再决定观看哪一部，也可以一次性观看多个影片，我们可以观察观看统计数据，例如某个影片是被完整观看还是只观看了一部分^。9

Streaming has not only changed the way our members interact with the service, but also the type of data available to use in our algorithms. For DVDs our goal is to help people fill their queue with titles to receive in the mail over the coming days and weeks; selection is distant in time from viewing, people select carefully because exchanging a DVD for another takes more than a day, and we get no feedback during viewing. For streaming members are looking for something great to watch right now; they can sample a few videos before settling on one, they can consume several in one session, and we can observe viewing statistics such as whether a video was watched fully or only partially.⁹

突然之间，文化现实泄露到了统计洁净室，算法只计算用户、电影、评分和时间戳。现在，Netflix 可以精准追踪用户观看特定节目的方式，他们在选项之间犹豫了多久，甚至可能追踪他们暂停、快进或倒带的次数。流媒体的即时满足感创造了一种不同的评分关系——不是我上周或十年前看过的电影的评价，而是现在……Netflix 不再基于评分构建电影之间抽象关系的模型，而是构建用户在其各种应用中的实时行为模型。

All of a sudden, cultural reality leaks into the statistical cleanroom where algorithms count nothing but users, movies, ratings, and timestamps. Now Netflix can track precisely how their customers watch particular shows, how long they hesitate between options, and perhaps even how much pausing, fast-forwarding, or rewinding goes on. The instant gratification of streaming creates a different kind of rating relationship—not the evaluation of a film I watched last week or ten years ago, but right now. Netflix is no longer constructing a model of abstract relationships between movies based on ratings, but a model of live user behavior in their various apps.

这篇文章还指出，目前已有数百种设备可以播放 Netflix 内容，每种设备都提供各自的上下文数据流。2013 年，Netflix 推出了其 Facebook 集成系统的美国版，允许用户根据 Facebook 好友的评分高低查看推荐内容（该功能后来停用，但用户仍可在社交媒体上分享他们正在观看的内容）。^所有这些新的数据流为该服务创建了更丰富的上下文，增加了数百个可能的数据点来影响推荐。

The post goes on to note the hundreds of devices that can now stream Netflix content, each providing its own streams of contextual data. And in 2013, Netflix unveiled the U.S. version of its Facebook integration system, allowing users to see recommendations based on what their friends on Facebook rated highly (this feature was later discontinued, though users can still share what they’re watching on social media).¹⁰ All of these new data streams create a much richer context for the service, adding hundreds of possible data points to inflect recommendations.

从邮寄DVD到数字流媒体的转变，彻底重塑了Netflix文化机器。流媒体的即时满足感重塑了Netflix的体验。我们不再是在一个拥有数十万种选择的庞大数字仓库中浏览，寻找几天或几周后才会喜欢的产品；而是浏览一个相对较小的目录，寻找可以立即开始观看的内容。正如Netflix工程师所说，流媒体网站“一切都是推荐”，该公司特意为用户提供线索，以便他们“了解我们是如何适应他们的口味的”。11^工程师们不仅在博客上分享他们的工作，而且界面本身也解释了为什么推荐某些特定内容（至少在表面上是这样）。选项的范围是完全量身定制的，类别和特定项目根据个人用户而浮动，并带有标签，例如“因为你喜欢纸牌屋”。与我们迄今为止讨论过的许多算法系统不同，Netflix 拉开了窗帘的一角，邀请用户一睹用于构建他们的文化机器的行为模型。

The shift from DVDs by mail to digital streaming was a complete reinvention of the Netflix culture machine. The instant gratification of streaming reframes the intellectual experience of Netflix. We are no longer browsing a vast digital warehouse with hundreds of thousands of options for a product we will only enjoy days or weeks later; we are surfing a comparatively smaller catalog, looking for something to start watching immediately. As the Netflix engineers note, with the streaming site “everything is a recommendation,” and the company deliberately seeks to clue in users so they can “be aware of how we are adapting to their tastes.”¹¹ Not only do the engineers blog about their work, but the interface itself offers explanations for why particular things are recommended (at least on a superficial level). The universe of options is completely tailored, with categories and specific items floating to prominence based on the individual user and labeled with tags like “because you liked House of Cards.” Unlike many of the algorithmic systems we have discussed so far, Netflix pulls back a corner of the curtain, inviting users to catch glimpses of the behavior model used to build their culture machine.

算法理念的巨变也促使Netflix重新审视其整个视频内容策略。如果说算法1.0模型将其库中的每一部电影和电视剧都视为一种文化黑匣子，其影响力只能通过用户评分来衡量，那么算法2.0模型则从研究内容本身开始：好莱坞和全球电影产业的庞大产出、授权协议、类型片预期、明星影响力以及诸多其他因素。正如负责此次改革的高管托德·耶林所说：“我的首要目标是：拆解内容！” ¹²该公司开始着手构建新的创意本体，通过识别内容的构成要素及其背后的模式。

The sea change in algorithmic philosophy also led Netflix to reconsider its entire approach to video content. If the algorithm 1.0 model treated each film and television show in its library as a kind of cultural black box whose impact could be measured only through user ratings, the algorithm 2.0 model began by studying the catalog itself: the sprawling output of the Hollywood and global film industries, licensing agreements, genre expectations, star power, and many other factors. As Todd Yellin, the executive in charge of the overhaul, put it, “My first goal was: tear apart content!”¹² The company set out on a path to consruct a new creative ontology by identifying the building blocks of their content and the patterns behind them.

亚历克西斯·马德里加尔 (Alexis Madrigal) 和伊恩·博格斯特 (Ian Bogost) 在《大西洋月刊》上发表了对耶林作品的广泛分析，发掘出一个非凡的新算法系统，该系统将 Netflix 视频库的复杂多样性变成了一门科学。13^耶林描述了在向网络视频转型后推动 Netflix 分析的新分析哲学，他半开玩笑地将这个想法称为 Netflix 量子理论。该公司决定直接测量电影和电视节目，只要有足够多的人工标记者和一份经过精心调整的三十六页指南来测量大约 1,000 个量子或微标签（耶林现在更喜欢这样称呼它们），这些量子或微标签将作品置于类型预期的系统景观中。14通过追踪每部电影或节目中的数^十个变量，包括脏话程度、女性角色的实力以及结果的模糊性或确定性，Netflix 组建了一个复杂的算法模型来描述单个电影和电视作品之间的文化关系，该模型完全涵盖了计算和文化之间的差距。

Alexis Madrigal and Ian Bogost published an extensive analysis of Yellin’s work in The Atlantic, unearthing a remarkable new algorithmic system that made a science of the complex diversity of Netflix’s video library.¹³ Yellin described the new philosophy of analysis that drove Netflix analytics after the transition to web video, an idea he half-jokingly called Netflix Quantum Theory. The company decided that movies and television shows could be measured directly, given enough human taggers and a finely tuned thirty-six-page guide to measuring about 1,000 quanta, or microtags, as Yellin now prefers to call them, which position the works within a systemic landscape of genre expectations.¹⁴ By tracking dozens of variables in each film or show, including the level of profanity, the strength of female characters, and the ambiguity or certainty of the outcome, Netflix has assembled a sophisticated algorithmic model for describing the cultural relationships among individual film and television works, a model that fully embraces the gap between computation and culture.

Netflix 所做的是彻底重新划分有效可计算性的界限，放弃了其最初的论点，即通过严格约束的行为（五分制评分、排队等）可以构建消费者观影欲望的最佳模型。新的论点并非倒退，而是在扩展可计算领域的一次雄心勃勃的飞跃，它宣称复杂的文化概念，例如电影幽默的明暗程度，是可以量化的。支撑这条扩展战线的“群兽”是人类：经过训练的个体，被要求参与离散的文化计算行为，例如在简单的数值尺度上评估电影的模糊性。N. Katherine Hayles 将《我的母亲是一台计算机》命名为双关语，因为“计算机”一词最初指的是进行计算的人，通常是女性。如今，“标记器”一词在计算语境中经历了相反的旅程。它最初是指对文本主体执行标记的计算机程序（例如，词性语言标记器），但现在它是 Netflix 的职位名称。

What Netflix has done is completely redraw the lines of effective computability, abandoning its first thesis that tightly constrained behaviors (rating on a five-point scale, putting things in a queue, etc.) could yield an optimal model of consumer viewing desire. The new thesis is not a retreat but an ambitious leap forward in expanding the calculable terrain, declaring that complex cultural concepts like the lightness or darkness of a film’s humor can be quantified. The pack animals supporting this expanded line of battle are human beings: trained individuals who are asked to engage in discrete acts of cultural computation, such as evaluating filmic ambiguity on a simple numerical scale. N. Katherine Hayles titled My Mother Was a Computer as a pun on the fact that the noun “computer” originally referred to a human, typically a woman, performing calculations. Now the word “tagger” has made the opposite journey in the context of computation. It originally referred to computer programs performing markup on a body of text (e.g., a parts-of-speech linguistic tagger), but now it is a job title at Netflix.

这些匿名的标注员让我们首次清晰地瞥见了人类劳动力在计算效率的幌子下支撑着一台文化机器，推动着黑箱的齿轮运转。Netflix 隐藏这些标注员，并非因为他们羞于用人取代机器，而是因为他们是这个黑箱不可或缺的一部分。这套混合计算系统，尤其是该公司长达36页的量子理论示意图的细节，都是宝贵的知识产权。这项工作本身笨拙地将娱乐媒体与泰勒因式分解法融合在一起，导致标注员格雷格·哈蒂不得不使用一些奇怪的连词来为自己和他人辩护这份工作的严肃性：“我知道这不是真正的工作。但我为此感到自豪，也为它的质量感到自豪。我牢记，我在工作，而不是偷懒。我的心态是，这仍然是工作，我坐在办公桌前。” ¹⁵哈蒂说，他的朋友自然希望他向他们推荐电影，但他却建议他们使用 Netflix。

These anonymous taggers offer one of our first clear glimpses of the human labor power supporting a culture machine behind the facade of computational efficiency, trundling the gears of the black box. Netflix hides the taggers away not because they are ashamed of replacing machines with humans, but because they are so much a part of that black box. This hybrid computational system, and especially the details of the company’s thirty-six-page schematic for its quantum theory, are valuable intellectual property. The job itself awkwardly blends entertainment media and Taylorist factorization, leading to the strange conjunctions that tagger Greg Harty must employ to defend the seriousness of his employment to himself and others: “I know this is not real work. But I’m proud of it and I’m proud of the quality. I keep in mind that I’m working and not slacking off. My mindset is that it’s still work and I’m at a desk.”¹⁵ Harty reports that his friends naturally expect him to recommend movies to them, but he urges them to use Netflix instead.

这句话暗示了算法2.0环境本质中一些有趣的地方。Netflix量子理论的机器人装置避免了机器中人类之间的直接人机交互，而是构建了一种机械的、统计性的视角来观察我们使用各种Netflix应用的行为。哈蒂永远不会突然出现在屏幕上，开始建议你接下来观看什么。然而，人类是文化与计算之间套利的不可或缺的一部分，他们以高度复杂但受限的方式评估视频。这与许多花费数百万美元向我们保证有真人随时准备帮助我们购买汽车保险或获得新抵押贷款的公司形成了鲜明对比。对于Netflix来说，品牌是算法的，魔法是计算的，而人类则被小心地置于聚光灯之外。

That remark hints at something interesting about the nature of this algorithm 2.0 environment. The cyborg apparatus of Netflix quantum theory avoids direct human engagement among the humans in the machine, structuring a mechanical, statistical gaze on our actions using the various Netflix apps. Harty is never going to pop up on-screen and start suggesting what you watch next. And yet the humans are integral to the arbitrage between culture and computation, evaluating video in highly sophisticated yet constrained ways. Consider this as a contrast to the many companies that spend millions assuring us that real human beings stand ready to help us buy car insurance or get a new mortgage. For Netflix, the brand is algorithmic, the magic computational, and the humans are kept carefully out of the spotlight.

该系统生成了一系列引人入胜的结果，即 Netflix 目录的知识本体，与 Google 的知识图谱不同，它更容易被外部检查。正如 Bogost 和 Madrigal 兴高采烈地报告的那样，Netflix 已经生成了一套类型类别，将好莱坞电梯游说中的概念三张牌蒙特卡罗（《电子情书》 + 《2001》 = 《她》）与严格算法的本体论完整性相结合。这个上层建筑为一个已经对此类问题着迷的行业提供了分析分类法。好莱坞的叙事学从亚里士多德延伸到神话学者约瑟夫·坎贝尔，这是一个备受争议的《塔木德》传统，它对《大白鲨》的影响力不亚于坎贝尔的《英雄之旅》；毕竟，叙事中的细微变化可以转化为数百万的票房收入。像《救猫咪》这样精心设计的剧本创作“系统”体现了好莱坞电影制作在^各个层面上错综复杂的规则和期望，从类型和背景到构成每个剧本叙事基石的个别“节奏” ^。17

The system generates a fascinating body of results, a knowledge ontology of the Netflix catalog that, unlike Google’s KnowledgeGraph, is much more permeable to external examination. As Bogost and Madrigal gleefully report, Netflix has generated a set of genre categories that combine the conceptual three-card monte of the Hollywood elevator pitch (You’ve Got Mail + 2001 = Her) with the ontological completism of a rigorous algorithm. This superstructure offers an analytical taxonomy of an industry that is already obsessive about such matters. Hollywood’s narratology stretches from Aristotle to scholar of myth Joseph Campbell, a fiercely contested Talmudic tradition that lends as much weight to Jaws as Campbell’s The Hero’s Journey; after all, minor inflections in narrative can translate into millions at the box office. Elaborate screenwriting “systems” like Save the Cat signal the intricate rules and expectations for Hollywood productions¹⁶ at every level, from genre and setting to the individual “beats” that make up the narrative building blocks of each script.¹⁷

Netflix 并未追踪这些故事中的每一个“节拍”，但马德里加尔和博格斯特发现，它已经精心构建了一个包含 76,897 个类型的图式，其中一些仍在等待电影或电视剧的填补。该算法揭示了好莱坞主流的刻板公式主义，同时也赞扬了排列组合的创造力，识别出三部符合“维多利亚时代英国喜剧”类别的作品，以及“德语二战情感电影”的空集。Netflix 图式中存在这些空集本身就意义非凡，表明了该系统所构建的元结构的复杂性，并由此延伸出对普遍知识渴望的表达。Netflix 机器正在触及那些可能存在的电影和类型：量子理论预测但尚未观测到的稀有娱乐粒子。

Netflix does not track each “beat” in these stories, but Madrigal and Bogost found that it has elaborated a schema of 76,897 genres, some of them still waiting for a film or TV show to fill them. The algorithm unveils the slavish formulism which dominates Hollywood but also celebrates the creativity of permutation, identifying the three works that qualify for the “British Comedies Set in the Victorian Era” category and the null set of “Emotional German-Language WWII Movies.” The existence of these null sets in the Netflix schema is itself remarkable, indicating the sophistication of the meta-structures the system has put in place and, by extension, an expression of the desire for universal knowledge. The Netflix machine is reaching for films and genres that might be: rare particles of entertainment predicted by their quantum theory but not yet observed.

该系统不仅仅是一个简单的新奇玩具（尽管马德里加尔和博格斯特显然很喜欢创建自动类别生成器来模仿它）。Netflix 使用这些微标签来驱动其“一切皆推荐”的设计，这意味着该算法不仅塑造个人选择，还塑造他们所出现菜单的框架。正如一位公司代表所说，“Netflix 有 3300 万个不同版本”，或者说为每个客户量身定制一个系统。18^该系统还允许将基本功能组合成更复杂的分类。例如，“感觉良好”这个描述词不是手工编码的，而是从其他标记的功能中衍生出来的，例如结局的幸福感和电影的整体喜剧评级。就像好莱坞大亨一样，Netflix 可以根据用户和内容制作者的行为得出新的排列组合。

The system is more than a simple novelty toy (though Madrigal and Bogost clearly enjoyed creating automated category generators to spoof it). Netflix uses these microtags to drive its “everything is a recommendation” design, meaning that the algorithm shapes not only the individual choices but the framing of the menus they appear in. As one company representative put it, “there are 33 million different versions of Netflix,” or a uniquely tailored system for each individual customer.¹⁸ The system also allows basic features to be combined into more sophisticated categorizations. The descriptor “feel-good,” for example, is not coded by hand but rather derived from other tagged features such as the happiness of the ending and overall comedic rating of the film. Like a Hollywood mogul, Netflix can derive new permutations based on the actions of its users and content producers.

这种对影视复杂创意领域的多变量分析也产生了意想不到的效果，比如佩里·梅森 (Perry Mason )在算法分类中神秘的突出地位。根据 Netflix 数据得出的顶级导演名单涵盖了所有常见的热门导演……以及《佩里·梅森》的导演克里斯蒂安·I·尼比二世 (Christian I. Nyby II)。Netflix 最受欢迎的演员名单包括布鲁斯·威利斯 (Bruce Willis) 和成龙，而《佩里·梅森》的主演雷蒙德·伯尔 (Raymond Burr) 位居榜首。当马德里加尔向耶林询问这一现象时，他的回答描述了一种我们才刚刚开始应对的与算法文化的新型批判关系：

This multivariable analysis of the complex creative field of film and television also generated its own unintended effects, such as the mysterious prominence of Perry Mason in the algorithmic categories. A list of top directors derived from Netflix data included all the usual suspects … and Perry Mason director Christian I. Nyby II. A list of Netflix’s favorite actors, including Bruce Willis and Jackie Chan, was topped by Raymond Burr, star of Perry Mason. When Madrigal questioned Yellin about the phenomenon, his response described a new kind of critical relationship with algorithmic culture that we are just beginning to grapple with:

让我稍微哲学一下。在人类世界里，生活因偶然性而变得有趣。你给机器世界增加的复杂性越多，你就越能想象到意外的惊喜。佩里·梅森即将发生。机器里的这些幽灵总是复杂性的副产品。有时我们称之为bug，有时我们称之为特性^。19

Let me get philosophical for a minute. In a human world, life is made interesting by serendipity. The more complexity you add to a machine world, you’re adding serendipity that you couldn’t imagine. Perry Mason is going to happen. These ghosts in the machine are always going to be a by-product of the complexity. And sometimes we call it a bug and sometimes we call it a feature.¹⁹

Netflix 构建 Cinematch 及其后续系统，旨在通过在正确的时间向我们提供正确的选项来制造意外的惊喜，就像 Google Now 一样，只是换了个名字来制造预期。但耶林在此论证的是，该系统创造的意外惊喜实际上并非由人类控制。策划该系统的总工程师抽象地理解其输出，却无法解释具体结果。佩里·梅森之谜虽然异想天开，却也说明了问题：这个精心调整的算法系统生成的结果无法通过其创建规则来理解，只能在其过程中观察到。文化机器的实施本身就带来了惊喜。

Netflix built Cinematch and its successor system to manufacture serendipity by presenting us with the right options at the right time, just like Google Now, crafting anticipation by another name. But what Yellin argues here is that the serendipity created by the system is not really under human control. The chief engineer who masterminded this system understands its outputs in the abstract but has no explanation for this specific outcome. The Perry Mason mystery is whimsical but telling: this fine-tuned algorithmic system generates results that can’t be understood by the rules of its creation, but only observed in its process. The implementation of the culture machine provides its own surprises.

这完美地诠释了魔法在计算中仍然发挥着弥合复杂系统中因果差距的作用。在层层抽象和流程中，即使是像耶林这样的工程师也无法理解Netflix算法的内部运作，我们却创造了一个隐喻的机会，一个象征性能动性的形式。20^耶林对幽灵的召唤，这个经典的隐喻失控的象征，巧妙地概括了这些抽象层的问题。尽管其算法2.0推荐模型直接应对了计算与文化之间的差距，但Netflix量子理论的逻辑结构与视频内容的制作和消费方式之间仍然存在着缝隙和脱节，有时佩里·梅森的形象就体现在了这一差距中。即使是隐喻机器的构建者，抽象概念的守护者，也无法解释特定效应产生的原因，因为抽象本身使得特定类型的知识变得不可见。作为人类，我们必须重新插入我们自己的意义结构、语义或概念粘合剂，以填补抽象层之间的空白，引入鬼魂、神灵和其他故事来解释这些谜团。

This is a beautiful illustration of the function that magic still plays in computation as a way to bridge causal gaps in complex systems. Somewhere in the layers of abstraction and process that insulate even an engineer like Yellin from the inner workings of the Netflix Algorithm, we create an opportunity for metaphor, for a form of symbolic agency.²⁰ Yellin’s invocation of the ghost, that classic symbol of the metaphor run wild, neatly encapsulates the problem with these layers of abstraction. Even though its algorithm 2.0 model for recommendations grapples directly with the gap between computation and culture, there are still seams and disjoints between the logical structures of Netflix Quantum Theory and the ways that video content is produced and consumed, and sometimes Perry Mason materializes in that gap. Even the architects of the metaphor machine, the keepers of the abstractions, cannot explain why particular effects occur, because the abstraction itself makes particular kinds of knowledge invisible. Being human, we must re-insert our own structures of meaning, of semantic or conceptual glue, to backfill the gaps between the layers of abstraction, bringing in ghosts, gods, and other stories to explain these mysteries.

与此同时，我们深深地被这些抽象的系统所吸引，被简洁界面和清晰本体的浪漫所吸引。即使Netflix的推荐系统倾注了数千小时的人力，它仍然呈现出一个天衣无缝的计算外观，因为我们已经到达了一个阶段：相比陌生人的建议，我们中的许多人更愿意相信一台陌生计算机的建议。推荐系统之所以如此成功，是因为它把判断任务置于黑盒之中，要求我们相信算法中嵌入的个性化效果。相比之下，阅读影评或浏览IMDb或烂番茄等网站，需要我们以一种更为复杂、更为人性化的方式去评价评估者，衡量其他可能与我们品味不同的人给出的建议的适用性。算法并没有表现出这种影响或个性的暗示——它声称自己是我们兴趣和欲望的一面镜子。引入另一个人类角色，会玷污这种修辞契约的纯粹性，要求我们不仅衡量自己和当下的欲望，还要衡量另一个可能知道也可能不知道我们想要什么的人。因此，哈蒂和他的同伴们，在大多数情况下，都必须隐藏在黑匣子里。

At the same time, we are deeply compelled by these abstracting systems, by the romance of clean interfaces and tidy ontologies. Even with thousands of human hours encoded into its recommendations, Netflix presents a seamless computational facade, because we have arrived at a stage where many of us will trust a strange computer’s suggestions more than we will trust a stranger’s. The rhetoric of the recommendation system is so successful because it black boxes the task of judgment, asking us to trust the efficacy of personalization embedded in the algorithm. By contrast, reading movie critics or browsing sites like IMDb or Rotten Tomatoes requires us to evaluate the evaluators in a much more complicated, human way, measuring the applicability of advice generated by other personalities who might not share our tastes. The algorithm presents no such affect or hint of personality—it alleges to be a mirror of our interests and desires. To introduce another human character into the equation would be to sully the purity of that rhetorical contract, asking us to measure not just ourselves and our desires in the moment but another person who might or might not know what we want. And so Harty and his fellow taggers must remain, for the most part, hidden in the black box.

Netflix 完全定制的前端让我们得以一窥算法时代的第二个主要特征：抽象美学。Uber、谷歌和亚马逊等公司正以某种特定的风格构建自己的帝国，这种简化的理念要求抽象出复杂而繁琐的细节，以提供可靠且持久的服务。这些公司从事着某种形式的算法套利，为我们处理繁琐的细节，并成为每笔交易的中间人。他们的角色最初类似于私人助理或总承包商，但逐渐演变成大维齐尔或土耳其政府的翻译官，不仅行使权力来执行我们的决策，还控制决策路径和代理空间。抽象经济依赖于能力、信任和开放的美学，以建立这种亲密的算法共享形式所需的那种融洽关系。我们将在第 4 章和第 5 章中回顾算法套利对劳动力和财务的影响，但首先我们需要将这种现象理解为一种文化立场和一种创造性实践。

The totally customized front-end of Netflix offers a glimpse at a second major trope of the algorithmic age: the aesthetics of abstraction. Companies like Uber, Google, and Amazon are building their empires on a particular style, an ethos of simplification that requires abstracting away complex and messy details in order to deliver a reliable and persistent set of services. These companies are engaged in a form of algorithmic arbitrage, handling the messy details for us and becoming middlemen in every transaction. The role begins as something like a personal assistant or a general contractor but gradually evolves into a position of grand vizier or dragoman of the Porte, exercising the power not merely to enact our decisions, but to control the decision pathways, the space of agency. The economies of abstraction depend on an aesthetics of competence, trust, and openness to build the kind of rapport that such intimate forms of algorithmic sharing require. We’ll return to the labor and financial implications of algorithmic arbitrage in chapters 4 and 5, but first we need to understand the phenomenon as a cultural position and a kind of creative practice.

个性化的艺术

The Art of Personalization

Netflix 成立于 1997 年，是首批将计算抽象作为商业计划的网络公司之一。视频市场很容易被打入，尤其是因为大型连锁店多年来一直忽视 Netflix：该公司鼓励顾客完全避开零售体验，在家中舒适地购买种类繁多的电影。如今，Netflix 已成为视频娱乐领域的主导提供商，收入开始与票房本身相媲美，在数字内容授权谈判中，Netflix 也面临着越来越大的风险。21该公司已开始寻找其他方式，以便在其虚拟货架上提供更多商品。与亚马逊、微软和其他数字流媒体服务一样，Netflix 现在委托制作自己的节目，以便提供独家内容，并征服在线流媒体视频的新领域。2016 年 1 月 6 日，首席执行官里德·哈斯廷斯宣布将在全球 130 个国家/地区推出流媒体服务，使该服务成为“新的全球互联网电视网络”。22^{哈斯廷斯在}发布会上的讲话直击抽象娱乐体验的所有亮点：

Founded in 1997, Netflix was one of the first dot-coms to use computational abstraction as a business plan. The video market was easy to break into, especially because the major chains ignored Netflix for years: the company invited customers to sidestep the retail experience entirely by shopping for a much wider selection of films from the comfort of your home. Now that it has become the dominant provider of video entertainment, starting to rival the box office itself for revenue, Netflix has seen the stakes grow in licensing negotiations for digital content.²¹ The company has started to look for alternative ways to get more items on its virtual shelves. Like Amazon, Microsoft, and other digital streaming services, Netflix now commissions its own shows so that it can offer exclusive content and conquer the new arena of online streaming video. On January 6, 2016, CEO Reed Hastings announced the launch of global streaming to 130 countries, making the service a “new global Internet TV network.”²² Hastings’s words at the announcement hit all the highlights of an abstracted entertainment experience:

随着这项服务的推出，从新加坡到圣彼得堡，从旧金山到圣保罗，世界各地的消费者将能够同时观看电视节目和电影，无需等待。借助互联网，我们将权力交到消费者手中，让他们能够随时随地使用任何设备观看^。23

With this launch, consumers around the world—from Singapore to St. Petersburg, from San Francisco to Sao Paulo—will be able to enjoy TV shows and movies simultaneously—no more waiting. With the help of the Internet, we are putting power in consumers’ hands to watch whenever, wherever and on whatever device.²³

无需等待，也不再受地域限制。从纸质邮件的革命性红色，到公司“随时随地、任何设备”的得意之作，Netflix 凭借抽象美学参与竞争并最终获胜。因此，当这家公司开始从亚马逊式的物流和供应链管理巨头（在全国范围内运输 DVD 方面比任何人都做得更好）转型为内容创造者时，它仍然依赖算法分析，这应该不足为奇。但是，当算法被用来塑造创作过程，创造出人类比机器更容易理解的公共文化作品时，会发生什么呢？

No more waiting, and no more geographical boundaries. From the revolutionary red of its paper mailers to the company’s triumphant announcement of service “whenever, wherever and on whatever device,” Netflix has competed and won based on the aesthetics of abstraction. It should come as little surprise, then, that when the company began to evolve from an Amazon-like titan of logistics and supply-chain management (moving DVDs around the country better than anyone else) to content creator, it would continue to depend on algorithmic analysis. But what happens when algorithms are used to shape the creative process, to create public works of culture that are far more legible to humans than to machines?

Netflix 委托制作了热门剧集《纸牌屋》，该剧于 2013 年首播，其制作很大程度上基于算法微积分：它有大量统计证据表明其用户会欢迎由凯文·史派西主演、大卫·芬奇执导的 BBC 政治剧的重启。24^为了在原创内容制作方面超越 HBO，该公司出价 1 亿美元获得了《纸牌屋》两季（共 13 集）的版权，使其成为电视（或“网络电视”，取决于你的定义）上最昂贵的剧集。25^与传统的试播模式不同，该公司投资了文化垄断（现在随着该节目获得康卡斯特等宿敌的授权，垄断地位正在崩溃）以拥有独特的创意产品。²⁶在圣丹斯电影节上，当该公司首席内容官泰德·萨兰多斯被问及算法对决策过程到底有多重要时，他回答说，决策过程中 70% 取决于数据，30% 取决于人类判断，“但如果说得通的话，那 30% 应该放在第一位。” ²⁷

Netflix commissioned its hit series House of Cards, which premiered in 2013, based in large part on algorithmic calculus: it had significant statistical evidence to suggest that its users would embrace a reboot of a BBC political drama starring Kevin Spacey, with director David Fincher at the helm.²⁴ Eager to outpace HBO as a producer of original content, the company bid $100 million to secure the rights to House of Cards for two thirteen-episode seasons, making it the most expensive drama on television (or “Internet TV,” depending on your definition).²⁵ Unlike a traditional pilot model, the company invested in a cultural monopoly (now crumbling as the show is licensed by arch-enemies like Comcast) to own a unique creative offering.²⁶ When the company’s chief content officer, Ted Sarandos, was questioned at the Sundance Film Festival about just how significant algorithms are to the decision process, he responded that it’s a 70 percent data, 30 percent human judgment mix, “but the thirty needs to be on top, if that makes sense.”²⁷

对于更传统的电视网络来说，这看起来像是一场豪赌，因为一部剧要与同一时段的其他几部剧竞争。但 Netflix 的算计则截然不同，尤其是当我们记住“一切都是推荐”这句话的时候。Netflix 知道它不必花费数百万美元来宣传这部剧，因为它已经可以直接接触到它的数百万用户。该公司用十个高度针对性的预告片来推广《纸牌屋》 ：为凯文·史派西的粉丝准备了凯文·史派西的预告片，为大卫·芬奇的粉丝准备了艺术镜头，并为刚刚看过像《末路狂花》这样以女性为主角的作品的观众准备了以女性角色为主角的场景。28 《纸牌屋》^不仅 是在最初的框架上，而且在制作和推出时都是算法制作的。

This looked like an immense gamble to more traditional television networks, where a show is competing against a few other offerings in the same time slot. But the calculation for Netflix is very different, particularly when we remember that “everything is a recommendation.” Netflix knew it did not have to spend millions advertising the show because it already has a direct line to its millions of users. The company promoted House of Cards to them with ten highly targeted trailers: Kevin Spacey for the Spacey fans, artful shots for the David Fincher fans, and scenes featuring female characters for viewers who had just seen something with strong female leads, like Thelma and Louise.²⁸ House of Cards was an algorithmically produced show not just in its initial framing but in its production and rollout.

换句话说，Netflix 确信《纸牌屋》会大获成功，首先是因为其初始要素，即可以根据量子理论衡量的公式成分，其次是因为它具有吸引客户注意力和兴趣的巨大力量。但对于算法模型来说，制作节目本身的细节相对不重要。从标记器的角度来看，合同上的墨水一干，最重要的部分就已经到位了：芬奇在，斯佩西在。让艺术家制作两季，Netflix 会确保我们观看。芬奇和该剧的其他导演对“标准和惯例”的缺乏感到惊讶，Netflix 赋予他们按照自己意愿创作节目的完全自主权，以及两季保证，这让他们能够以业内极不寻常的确定性来策划叙事。在最初决定投资《纸牌屋》之后，Netflix 开始使用算法来微观管理发行，而不是制作。就像那些算法股票交易公司一样，只要能够预测商品价格波动，很少关注其内在价值，Netflix 也知道，凭借人才与投资的结合，其自身的交付系统能够取得成功。或许颇具讽刺意味的是，这种抽象形式支持了非常人性化的导演模式，让芬奇和其他参与该剧的电影制作人能够自由地实现自己的审美愿景。29^在某种程度上，他们是那 30% 的顶层人物，推动着一系列由算法建模界定和实现的创意决策。

In other words, Netflix was confident that House of Cards would be a hit first because of its initial ingredients, namely the formula components it could measure according to Quantum Theory, and second because of its immense power to capture the attention and interest of its customers. But the details of producing the show itself were relatively unimportant for the algorithmic model. From a tagger’s perspective, the most important pieces were in place as soon as the ink dried on the contract: Fincher was there; Spacey was there. Let the artists make two seasons and Netflix would make sure we watched them. Fincher and the show’s other directors marveled at the absence of “standards and practices,” at the total autonomy Netflix granted them to create the show as they wished, and the two-season guarantee that allowed them to plot out the narrative with a degree of certainty highly unusual in the industry. After making the initial decision to invest in House of Cards, Netflix was using algorithms to micromanage distribution, not production. Like the algorithmic stock-trading firms that pay little attention to the intrinsic value of a commodity so long as they can predict the fluctuations in its price, Netflix knew that with this combination of talent and investment, its own delivery system could generate success. Perhaps ironically, this form of abstraction supports the very human auteur model, allowing Fincher and other filmmakers involved in the show free rein to achieve their own aesthetic visions.²⁹ In part, they were the 30 percent on top, driving a series of creative decisions that were scoped and enabled by algorithmic modeling.

这种全新创作模式最有力的信号是该剧的播出安排：全部十三集将于午夜上线，供观众在线观看。正如芬奇所说：

The most powerful signal of this new creative mode was the release schedule of the show, with all thirteen episodes appearing online for streaming at midnight. As Fincher put it:

周二晚上七点半的世界，已经死了。一根木桩刺穿了它的心脏，它的头被砍掉，它的嘴里塞满了大蒜。被俘虏的观众已经消失了。如果你给人们这样一个机会，让他们一天之内就能吸完所有毒品，我们有理由相信他们会这么做^。30

The world of 7:30 on Tuesday nights, that’s dead. A stake has been driven through its heart, its head has been cut off, and its mouth has been stuffed with garlic. The captive audience is gone. If you give people this opportunity to mainline all in one day, there’s reason to believe they will do it.³⁰

商业电视时间战场的这种新转变，是Netflix对抗传统零售模式漫长征程的逻辑延伸——始于在百视达排队的时间，终于在广告前等待预定播出的时间。这又一次美学上的抽象，削弱了更为传统的文化机器，突显了广告泛滥的播出时间表因其缺失而产生的侵扰性和决定性。一次性发布整季的举措也让Netflix有机会通过观察用户如何应对这种“大快朵颐”高质量内容的诱惑来检验这一假设。第一季发布时，一位用户在剧集上线后立即观看了全部13集，期间仅停顿了3分钟。31^第二季于周五晚发布，截至周末，2%的美国订阅用户已经观看了全部13集^。32

This new shift in the temporal battleground of commercial television was the logical extension of Netflix’s long march against traditional retail models—starting with the time spent in line at Blockbuster and concluding with the time spent in front of commercials waiting for a scheduled broadcast to begin. This is another aesthetic abstraction that undermined a more traditional culture machine, highlighting the intrusiveness and determinism of advertising-laden broadcast schedules through their absence. The move to release the full season at once also gave Netflix the opportunity to test the hypothesis by observing how users dealt with the invitation to gorge themselves on high quality content. When season 1 was released, one user watched all thirteen episodes immediately after they launched, pausing for only three minutes during the entire period.³¹ Season 2 was released on a Friday night, and 2 percent of all U.S. subscribers had watched the thirteen episodes by the end of the weekend.³²

正如萨兰多斯所说，在完善“粉丝的形成方式”的过程中，Netflix 构建了一种独特的时间美学，一种永恒的消费者当下，或称网络时间。33^该公司将好莱坞的创意制作与硅谷的分析技术相结合，打造了一种基于密集媒体消费（或称狂看）的艺术参与新模式。2015 年 9 月，该公司发布了一张图表，详细介绍了一项研究的结果，该研究分析了用户何时对某部剧集“上瘾”。他们将“上瘾”定义为 70% 的用户在观看完该季剧集后仍会继续观看的剧集。

In refining “how fans are made,” as Sarandos puts it, Netflix has constructed a distinct temporal aesthetic, a kind of eternal consumer present or network time.³³ The company deploys its combination of Hollywood creative production and Silicon Valley analytics to create a new model of artistic engagement based on intense bouts of media consumption, or binge-watching. In September 2015, the company released a graphic detailing the results of a study analyzing when users get “hooked” on a particular show, which they defined as the episode after which 70 percent of users stayed on through the end of the season.

“打造粉丝”的过程将创意与算法相结合，打造基于即时访问和完全定制的用户审美。Netflix 始终在线，并根据我们个人的审美历史——包括我们中途停播的节目、我们重温或快进的剧集和场景等等——快速推荐新的选择。这是一个永恒的消费者当下，一个不断受到算法影响的当下。

The process of “making fans” draws together creative and algorithmic production to craft a user aesthetic based on instant access and total customization. Netflix is always available, always quick to suggest new options based on our personal aesthetic histories—histories that include the shows we stopped halfway through, the episodes and scenes we rewatched or fast-forwarded through, and so forth. It is an eternal consumer present, a now constantly mediated by algorithms.

这种专注于创造特殊的个人时间，以“无限量观看”的方式发布高价值创意内容的做法，给 Netflix 观众带来了一系列新问题。Netflix 观众的理想化形象——一家人坐在家里，随时随地观看流媒体——与公司运营的网络接收空间相冲突，更不用说其白领消费者群体忙碌的生活了。在《纸牌屋》第二季首播时，Netflix 开发了一个“剧透屏蔽器”Twitter 过滤器，允许观众屏蔽媒体流中的剧透，直到他们“赶上”剧情。34^就连该剧的忠实粉丝奥巴马总统也在推特上呼吁社交媒体用户不要剧透剧情。35 Netflix 的“剧透屏蔽器”最初^是为另一部热播剧《绝命毒师》的欧洲首映而开发的，其删减言论既俏皮又直白地表明了自愿审查的立场（图 3.2）。要使用该应用程序，观众必须通过 spoilerfoiler.com 登录他们的 Twitter 帐户，这样 Netflix 就可以预先屏蔽可能泄露未观看剧集信息的推文。

This focus on creating special, individual time, on releasing highly valued creative content on an “all-you-can-watch” basis, created a new set of problems for Netflix viewers. The idealization of the Netflix viewing audience—a family sitting at home streaming media exactly when they wish to—clashes with the networked space of reception the company operates in, not to mention the hectic lives of its white collar consumer base. By the debut of House of Cards season 2, Netflix had developed a “Spoiler Foiler” Twitter filter to allow viewers to block spoilers in their media streams until they had “caught up.”³⁴ Even President Obama, a dedicated fan of the show, tweeted a plea for those on social media to refrain from spoiling the plot.³⁵ The Netflix Spoiler Foiler was originally developed for the European premiere of another hit show, Breaking Bad, and its rhetoric of redaction is both playful and a bald statement of voluntary censorship (figure 3.2). To use the app, viewers must log in to their Twitter accounts via spoilerfoiler.com, allowing Netflix to preemptively block potentially revealing tweets about unviewed episodes.

要想成为完美的Netflix消费者，享受不受时间限制的自由观看体验，就必须接受另一种限制，以便人为地保留一个特定的新奇空间。这种行为实际上与媒体学者亨利·詹金斯在《融合文化》中所描述的完全相反，后者指的是一群忠实的剧透粉丝社群，他们会试图在每一季新剧播出前泄露《幸存者》等剧集的制作秘密。Netflix并没有让我们成为拥有强大自主权的粉丝，拥有强烈的自主感，以至于威胁到剧集的作者完整性，而是要求我们为了追求更完美的审美体验而放弃自主权^。36

To become a perfect Netflix consumer, reveling in the freedom of temporally unconstrained viewing, one must embrace another constraint, in order to artificially preserve a particular space for novelty. This kind of behavior is effectively the opposite of what media scholar Henry Jenkins described in Convergence Culture as dedicated fan communities of spoilers, who would attempt to reveal production secrets for shows like Survivor before the scheduled airing of each new season. Instead of performing as empowered fans with a sense of agency so strong that it threatens the authorial integrity of the show, Netflix asks us to go out of our way to relinquish agency in the quest for a more perfect aesthetic experience.³⁶

即时流媒体娱乐的抽象、完全个性化目录的审美与电视作为广播媒体的初衷相冲突，因为电视创造了一个同时观看和参与的社群。正如Netflix在2014年4月向股东所言，“互联网电视正在取代线性电视。应用程序正在取代频道，遥控器正在消失，屏幕正在激增。” ³⁷这些屏幕和应用程序比它们正在取代的电视更加亲密。它们正越来越贴近我们的脸和身体，从房间的另一边到口袋、膝盖和手中。我们在更广泛的场合和时间使用它们，例如早晨通勤。更重要的是，它们更接近我们的算法自我，从尼尔森收视率的抽象观众统计数据到Netflix应用程序收集的详细个人观看历史记录。从广播到算法娱乐的转变不仅导致了内容的重塑，也导致了用户行为的重塑。《纸牌屋》以及该公司无数的其他产品都是一种影院即服务，一种随时随地可观看内容的订阅服务。在这个未来的愿景中，一系列参与者被排除在外：全国和地方有线电视公司、广告商、尼尔森收视率及其衡量的全国直播观众，甚至本地网络的附属新闻团队。打开“剧透陷阱”之后，剩下的就只有我们自己，一个原子化的观众群体，直接与算法互动，体验着完全定制化的娱乐内容库（图3.3）。它让我们能够更充分地接受Netflix所倡导的特定抽象概念，在自己的私人时间流中观看节目，而Netflix也在观看我们。正如该公司要求观众自己策划和过滤对《纸牌屋》的体验一样，它在节目制作中建立了自己的抽象美学，这种美学几乎既反映了该公司的情况，也反映了虚构的华盛顿特区的权力斗争。

The aesthetic of the abstract, totally personalized catalog of instant streaming entertainment conflicts with the original logic of television as a broadcast medium, one that created a community of simultaneous viewing and engagement. As Netflix put it to its shareholders in April 2014, “Internet TV is replacing linear TV. Apps are replacing channels, remote controls are disappearing, and screens are proliferating.”³⁷ These screens and apps are far more intimate than the televisions they are replacing. They are moving closer to our faces and bodies, from across the room to pockets, laps, and hands. We use them in a wider variety of settings and times, like the morning commute. More significantly, they are closer to our algorithmic selves, from the abstracted audience statistics of Nielsen ratings to detailed personal viewing histories collected by the Netflix app. The shift from broadcast to algorithmic entertainment leads to a reinvention not merely of content but of user behavior. House of Cards and the company’s countless other offerings are cinema as a service, a subscription for content available whenever and wherever we want it. In this vision of the future, a whole series of players has been banished from the field: national and local cable companies, advertisers, Nielsen ratings and the live national audience they measure, and even the local network affiliate news crew. What remains after we switch on our Spoiler Foiler is just us, an atomized viewing audience, interacting directly with the algorithm and experiencing a totally customized library of entertainment (figure 3.3). It allows us to more fully embrace the particular kind of abstraction Netflix is promoting, watching the show in our own private temporal stream, while Netflix watches us. Just as the company asks its viewers to curate and filter their own experience of House of Cards, it has established its own aesthetic of abstraction in the production of the show, an aesthetic that says almost as much about the company as it does about power struggles in a fictional Washington, DC.

纸牌屋的框架

Framing House of Cards

将Netflix本身解读为一系列算法、界面和话语，比解读该系统所生产的文化产品更有助于理解其作为文化机器的角色。但当我们将《纸牌屋》视为一部部分由算法驱动、为Netflix更广泛的抽象美学而设计和构建的作品时，这种企业化、计算化的创作模式在剧集本身中便有所体现。当然，我们不可能在该剧的美学特征与最初资助该剧的商业决策之间建立强有力的因果关系，因为该商业决策的一部分是授权芬奇和他的团队全权按照他们认为合适的方式创作该剧。尽管如此，我认为对该剧不同寻常的两季投资，以及该公司在创意管理方面采取的特别不干预的策略，都留下了深刻的印记。Netflix买下了这部剧，并围绕它构建了一个精心设计的推荐反馈循环，选择了兼具美学和经济效益的对话方式。这种反馈循环以及支撑它的密集品牌推广和广告宣传，也让许多人将该剧的创意叙事与Netflix联系起来。

Reading Netflix itself as a series of algorithms, interfaces, and discourses is far more instructive for understanding its role as a culture machine than reading the cultural products produced by the system. But when we consider House of Cards as a creation that is in part algorithmic, designed for and structured within the broader abstraction aesthetic of Netflix, particular traces of this corporate, computational authorship emerge in the show itself. It is impossible, of course, to draw a strong causal link between aesthetic features of the show and the business decision that first funded it, because part of that business decision was to grant Fincher and his team carte blanche to create the show as they saw fit. Nevertheless I will argue that the unusual two-season investment in the show, and the particularly hands-off approach the company has taken to creative management, have left their mark. Netflix bought this show and created an elaborate recommendations feedback loop around it, choosing the terms of a conversation that is both aesthetic and financial. That feedback loop and the intensive branding and advertising that support it have also led many people to associate the creative narrative of the show with Netflix.

萨兰多斯再次表示：“我尽量不去选择那些定义我们的节目。我们的品牌真正关乎个性化。品牌就是找到你喜欢的、但在其他任何地方都找不到的东西。” ³⁸ Netflix 最终希望其消费者喜爱的不仅仅是内容，而是 Netflix 本身：应用程序、服务、平台。萨兰多斯指的是 Wendy Hui Kyong Chun 所理解的软件品牌：一个隐喻或工具包。Netflix 的标志代表着即时访问经过算法过滤并专门为你策划的娱乐菜单。对许多消费者来说，《纸牌屋》是他们接触这个品牌的起点，这部剧在他们如何解读更广泛的 Netflix 体验方面发挥着重要作用。正如我们将看到的，这部剧概括了其文化容器的抽象美感，为观众创造了独特的影响或情感记录。该节目的框架，无论是从 Netflix 的投资理由还是向观众介绍该节目的方式来看，都揭示了计算主义系统与其个性化修辞之间的一系列紧张关系，以及无限流媒体内容的承诺与为独家媒体产品精心培养观众之间的一系列紧张关系。

Sarandos again: “I’m trying not to pick shows that define us. Our brand is really about personalization. The brand is finding the thing you love that you can’t find anywhere else.”³⁸ The thing that Netflix ultimately wants its consumers to love is not just the content but Netflix itself: the application, the service, the platform. The thing Sarandos means is a software brand in Wendy Hui Kyong Chun’s sense of the term: a metaphor or toolkit. The Netflix emblem stands for instant access to a menu of algorithmically filtered entertainment specially curated for you. For many consumers, House of Cards was their introduction to this brand, and the series plays an important role in how they read the broader Netflix experience. As we’ll see, the show recapitulates the abstract aesthetic of its cultural container, creating a distinctive affect or emotional register for viewers. The framing of the show, both in terms of Netflix’s justifications for its investment and the ways that the program was introduced to viewers, reveals its own set of tensions between the computationalist system and its rhetoric of personalization, and between the promise of infinite streaming content and the careful cultivation of audiences for an exclusive media product.

《纸牌屋》的片头字幕体现了该剧的美学理念，而 Netflix 平台不鼓励跳过这些字幕，巧妙地强调了这一理念。39^这些场景是精美的延时 HDR（高动态范围）摄影：需要自动相机将多张照片合成一张合成图像。它们以街景和华盛顿特区地标为特色，例如肯尼迪中心和联合车站（以及一些来自该剧拍摄地点巴尔的摩的镜头），车流在视野中快速穿梭。每个场景中都没有一个人。这是一种抽象的城市观，将城市视为一个政治和信息权力中心，强调机构而非人（图 3.4）。

The opening credits of House of Cards signal the show’s aesthetic philosophy, and that message is subtly underlined by the fact that the Netflix platform discourages skipping over them.³⁹ The scenes are beautifully rendered time lapse HDR (high dynamic range) photography: images that require automated cameras compositing multiple shots together into a single, synthetic image. They feature street scenes and DC landmarks such as the Kennedy Center and Union Station (and some shots from the show’s filming location, Baltimore) with accelerated traffic zooming in and out of view. Absent from every single scene is a single human being. It is an abstracted view of the city as a political and informational power center that emphasizes institutions over people (figure 3.4).

负责制作片尾字幕的摄影师将这项工作描述为“以一种肮脏、粗糙、邋遢的方式展现华盛顿特区” ^。40但我们眼前的城市却是禁欲主义的、流线型的、抽象的：这是系统本身的美学可视化。图像以静止的长镜头展现了汽车和办公室灯光、建筑和结构、天气和运动模式的数据点。它呈现了国会大厦简朴庄严的景象，同时也是一种算法视角，人类只能通过其物理和文化痕迹才能被看到。这就像一首关于实现的视觉诗，一件精雕细琢的数字艺术作品，向我们展示了一个抽象的物质宇宙的面貌。空旷的广场和公园以超现实的方式呈现了计算与文化之间的差距，并运用高动态范围成像技术更忠实地呈现了它们的空虚。我们可以在控制论的背景下解读片尾字幕，将城市视为一个信息系统，其中的汽车灯光提醒我们原子像比特一样流动。这也是对象征性魔力的庆祝，强调政治和社会结构的永恒稳固性，因为无形的人性浪潮席卷了它们。

The cinematographer who created the credits described the job as “showing DC in a dirty, gritty, grungy way.”⁴⁰ But the city we see is ascetic, streamlined, abstracted: this is the aesthetic visualization of the system itself. The images show the data points of cars and office lights, buildings and structures, weather and movement patterns in long, unmoving chiaroscuro shots. It presents an austere, stately vision of the capitol, but also an algorithmic view of it, where human beings are visible only through their physical and cultural traces. This is a kind of visual poem to implementation, a finely wrought piece of digital art that shows us what an abstracted material universe looks like. The empty plazas and parks render the gap between computation and culture hyper-realistically, using HDR techniques to more faithfully represent their emptiness. We can read the credits in the context of cybernetics, with the city as an information system where car lights remind us that atoms flow like bits. It is also a celebration of symbolic magic, highlighting the eternal solidity of political and social structures as waves of invisible humanity wash over them.

片尾字幕也是对技术的鲜明诠释。哲学家尤尔根·哈贝马斯拓展了这一思想传统，区分了“系统”和“生活世界”。哈贝马斯认为，生活世界是一系列自然实践，如今与司法、官僚体系，尤其是资本主义力量的体系存在着紧张关系，这些力量试图规范、货币化和消除人类行为的某些方面。41^哈贝马斯的这一批判源于他对技术的解读，这种解读与吉尔伯特·西蒙东和伯纳德·斯蒂格勒的观点有许多相似之处，他也在同一个核心问题上挣扎，即在日益算法化的文化中理解人类的未来。42回想一下在屏幕的光芒下，Netflix 家庭的典型形象（图 3.3 ），这些^片尾字幕似乎邀请我们进入一场并非关于人类角色的演出，而是关于文化机器本身的演出，在这里，文化机器体现在同样不人道的国会政治机器中：一个关于生命与结构斗争的故事。

The credits also serve as a striking illustration of technics. The philosopher Jürgen Habermas extends that intellectual tradition to draw a distinction between “system” and “lifeworld.” Habermas conceives of the lifeworld as a set of natural practices that now exists in tension with the system of judicial, bureaucratic, and, especially, capitalistic forces that seek to regulate, monetize, and dispel certain aspects of human behavior.⁴¹ Habermas builds this critique from a reading of technics that has many parallels to Gilbert Simondon and Bernard Stiegler, and struggles with the same central issue of understanding humanity’s future in an increasingly algorithmic culture.⁴² Thinking back to the image of the prototypical Netflix family lit by the glow of their screens (figure 3.3), these credits seem to invite us into a show not about human characters but about the culture machine itself, embodied here in the equally inhuman political machine of the capitol: a story about the struggle between life and structure.

《纸牌屋》展现了这种框架美学，它描绘了一个充满抽象情感的世界，人物如同更大势力的傀儡和空壳。剧中真人演员的表演也印证了片尾字幕的抽象美学。故事追溯了参议员弗兰克·安德伍德（凯文·史派西饰）的阴谋诡计，他是一位雄心勃勃、经验丰富的政客。他和妻子克莱尔（罗宾·怀特饰）的婚姻生活既没有子女，又被严加管束，既有麦克白式的刻板印象，又透着不屈不挠的野心。弗兰克是那种在第一集中，邻居的狗被车撞了后，冷漠地掐死它的男人，但他和克莱尔似乎从未真正占据过这部剧的情感核心。他们的关键决策总是回避深刻的情感流露，而是依赖于一种冷漠而刻意的意图，这与片尾字幕中描绘的国会大厦毫无生气的街道遥相呼应。安德伍德一家是反英雄，是冷酷无情的典范，但他们对权力的贪婪也在某种程度上削弱了他们作为人物的魅力。《纸牌屋》与其说是关于他们，不如说是关于所有在政府和势力的庞大机器中挣扎的渺小生命。正如弗兰克在第一季结局中跪在祭坛前所说：“天上地下都没有慰藉。只有我们自己。渺小、孤独、奋斗、相互斗争。我向自己祈祷，为自己祈祷。” ⁴³

House of Cards unfolds this framing aesthetic with a world of abstracted emotions, of characters functioning as shells and figureheads for larger forces. The affect of the program’s human performers bears out the abstraction aesthetic of the credits. The narrative traces the machinations of Senator Frank Underwood (Kevin Spacey), a veteran political operator with aspirations to the presidency. He and his wife Claire (Robin Wright) share a childless, tightly controlled marriage with undertones of Macbeth and overtones of relentless ambition. Frank is the kind of man who dispassionately strangles a neighbor’s dog after it gets hit by a car in the first episode, and yet he and Claire never quite seem to claim the emotional heart of the show. Their pivotal decisions always pull back from deep revelations of emotion, relying instead on a kind of frosty, deliberate intention that echoes the soulless streets of the capitol portrayed in the credits. The Underwoods are antiheroes, paragons of ruthlessness, but also somehow lessened as characters by the urgency of their greed for power. House of Cards is not so much about them as it is about all of the small creatures struggling within the giant machines of government and influence. As Frank puts it in the season 1 finale, kneeling before an altar, “There is no solace above or below. Only us. Small, solitary, striving, battling one another. I pray to myself, for myself.”⁴³

这一幕只是该剧最引人注目之处的一个例子——芬奇选择让斯派西直接对着镜头说话，突然跨越了第四面墙，让我们作为演员融入到他的戏剧中。在柔和的色彩、抽象的政治和被束缚的人群的背景下，这一选择尤其引人注目，突然将图3.3中那个典型的家庭客厅融入到华盛顿权力游戏的版图中。就像Netflix本身一样，安德伍德的核心观众就是你，那个他经常与之目光接触的个体观众。个性化的神话已臻完美：Netflix为每位用户量身定制服务，但却用定制广告将数百万用户吸引到其高预算的创意实验中。我选择了这部剧，确信它经过算法计算符合我的口味，然后安德伍德亲自与我交谈，这是一种与每位新观众都营造出的亲密感。44这是一个老套^的伎俩，但它有力地唤起了一些新奇的东西：算法透过我们发光的屏幕回望着我们，目光坚定。斯派西的眼神交流看似私人，却经过精心设计，且可替换，正如他角色在剧中其他部分的关注一样。他跨越了故事的叙事界限，以一种既慈祥又精于算计的方式触动我们。

That scene is just one example of the show’s most remarkable feature—Fincher’s choice to allow Spacey to speak directly to the camera, suddenly leaping over the fourth wall to involve us as players in his drama. The choice is particularly arresting given this backdrop of muted colors, abstracted politics, and tight-leashed humans, suddenly incorporating that prototypical family living room from figure 3.3 into the geography of Washington power plays. Like Netflix itself, Underwood’s core audience is you, the individual viewer with whom he makes regular eye contact. The mythos of personalization is complete: Netflix customizes its offerings for every single user but drives millions of them to its high-budget creative experiment with tailored advertising. I choose the show, confident in its algorithmically calculated fitness for my tastes, and then Underwood addresses me personally, an illusion of intimacy that is performed with each new viewer.⁴⁴ It’s an old trick, but it powerfully evokes something novel: the steady gaze of the algorithm looking back through our glowing screens. Spacey’s eye contact seems personal, but it is calculated and fungible, just as his character’s attentions are in the rest of the show. He leaps over the diegetic boundary of the story to touch us in a way that manages to be both avuncular and calculating.

让《纸牌屋》中那个阴谋家、马基雅维利主义的头目直接与我们对话，为美学增添了一张面孔——史派西就是《纸牌屋》，他突然意识到了自己在故事中的位置。这一举动为安德伍德在华盛顿的阴谋增添了另一种愤世嫉俗的色彩，将他描绘成一个精明的表演者，通过伪装来获得权力，并强调了政治舞台的概念。我们被要求依靠他——他要求我们依靠他——来获得这部剧的道德影响力，就像安德伍德依靠自己一样。但这种联系也反过来起作用，将安德伍德表演中玩世不恭的算计带入《纸牌屋》本身的算法框架中，使这部剧成为一种精心制作的甜点，一种精心设计的诡计，让人不禁将这部剧如何在网上聚集观众与安德伍德如何在国会山操纵选票进行比较。

To have the machinator-in-chief, the lead Machiavellian in House of Cards’s rogues gallery, speaking directly to us puts a face to the aesthetic—Spacey is House of Cards, suddenly aware of his own diegetic position. The move adds another shade of cynicism to Underwood’s plotting in Washington, presenting him as a knowing performer dissimulating his way to power and underlining the concept of political theater. We are asked to depend on him—he asks us to depend on him—for the moral impact of the show just as Underwood depends on himself. But the linkage also works the other way, bringing the cynical calculation of Underwood’s performance into the algorithmic framing of House of Cards itself, making the show a kind of knowing confection, a constructed artifice that invites a comparison between how the series assembles its audience online and how Underwood whips votes on Capitol Hill.

因此， 《纸牌屋》体现了算法时代最诱人的神话之一：个性化的理想，为我们每个人特别定制内容。事实上，我们关心的内容，或者至少是昂贵且富有美感的内容，比如芬奇的节目，仍然相当有限。《纸牌屋》只有一部，但推广该剧的方式与Netflix的目标观众一样多。这就是信息理论家克里斯蒂安·桑德维格（Christian Sandvig）所说的“腐败的个性化”（corrupt personalization），此说法源于法律学者C·埃德温·贝克（C. Edwin Baker）和哈贝马斯（Habermas）：算法文化模糊了我们真正兴趣与一系列可能真正相关也可能不相关的商品之间的界限，例如我们的朋友在Facebook上“点赞”的产品，即使他们并非有意推荐。45^贝克和桑德维格最终都在此与哈贝马斯的生活世界概念相契合，指出了系统可以殖民或重新格式化生活世界的方式，将有机的、无方向的活动重组为人工管理以产生特定结果的活动。《纸牌屋》在一定程度上依赖于将该剧人为植入数百万 Netflix 用户的推荐和菜单中，并为每个用户提供量身定制的个性化建议。

House of Cards thus embodies one of the most seductive myths of the algorithmic age: the ideal of personalization, of bespoke content assembled especially for each one of us. In fact, the content, or at least the costly, aesthetically rich content we care about, like Fincher’s show, is still fairly limited. There is only one House of Cards, but there are as many ways to market the show as there are target Netflix viewers. This is what information theorist Christian Sandvig calls “corrupt personalization,” after legal scholar C. Edwin Baker and Habermas: the ways that algorithmic culture blurs the lines between our genuine interests and a set of commodities that may or may not be genuinely relevant, such as products “liked” by our friends on Facebook even if they did not knowingly endorse them.⁴⁵ Both Baker and Sandvig ultimately engage here with Habermas’s conception of lifeworld, identifying ways in which the system can colonize or reformat the lifeworld, restructuring organic, undirected activities into ones that are artificially managed to produce certain results. House of Cards depends, in part, on the artificial seeding of the show into the recommendations and menus of millions of Netflix users, to each of whom it is presented as a tailored, individual suggestion.

这种紧张关系是衡量实施差距的另一种标准，一方面是坐在客厅沙发上的观众对新剧着迷，另一方面是安德伍德作为 Netflix 算法关注的审美化身的凝视。Netflix 自信地将两季赌注押在《纸牌屋》上，因为它对这种共生关系有着深刻的统计学理解，它相信通过定制参考框架和数百万客户的推荐驱动界面，可以为该剧吸引粉丝。与所有电视和电影公司一样，Netflix 通过精心管理新闻稿和其他企业沟通工具（例如，剧透器）为《纸牌屋》制造文化意识和话语。但该公司还可以依靠其推荐算法和界面的计算权威，更进一步进行“腐败的个性化”，推荐该剧而不是向某个细分市场做广告。从公司的角度来看，这整套系统不过是又一次抽象的练习，一种将Netflix数百万个不同版本视为一个有解空间的算法问题的思考方式：如何才能让产品足够个性化，同时又能满足业务需求、授权机会和现有的创意内容库存？正如一位前产品工程副总裁所说：“Netflix追求最高效的内容。这里的高效是指，每花费一美元就能带来最大幸福感的内容。” ⁴⁶

This tension is another measure of the implementation gap, one spanned by viewers on the living room couch getting hooked on a new show, in one direction, and by Underwood’s gaze as the aesthetic personification of Netflix’s algorithmic attention for us, in the other. Netflix confidently placed its two-season bet on House of Cards because of its deep statistical understanding of this symbiotic relationship, its confidence that it could make fans for this show by tailoring the frames of reference, the recommendation-driven interfaces of its millions of customers. Like all television and film companies, Netflix manufactures cultural awareness and discourse for House of Cards through the careful management of press releases and other tools of corporate communication (e.g., the Spoiler Foiler). But the company can also rely on the computational authority of its recommendation algorithms and interface to go a step further with “corrupt personalization,” recommending the show rather than advertising it to a market segment of one. From the company’s point of view, this entire apparatus is simply another exercise in abstraction, a way of thinking about the millions of different versions of Netflix as an algorithmic problem with a solution space: how can the offerings be personal enough while fitting the business needs, licensing opportunities and creative content inventory on hand? As a former vice president of product engineering put it, “Netflix seeks the most efficient content. Efficient here meaning content that will achieve the maximum happiness per dollar spent.”⁴⁶

解读Netflix广告、界面乃至创意产品中的抽象美学，必然是一种非常受限的批判模式，我们必须认识到，这些解读只能带我们走这么远。营销和美学生产的空间如同计算海洋中的一条狭窄海岸线，受到Netflix算法以及许多其他文化力量的制约。Netflix文化机器的部分工作就是不断地在这片狭窄的美学海岸和其背后浩瀚的抽象海洋之间进行航向修正。这片海洋充斥着数十亿的数据点和概率推论，即使对于像托德·耶林这样的系统架构师来说，解读起来也更加困难。

Reading the aesthetics of abstraction in the advertisements, interfaces, and even creative products put out by Netflix is necessarily a very constrained mode of critique, and we have to recognize that these readings can take us only so far. The space of marketing and aesthetic production is a narrow shoreline on the ocean of computation, one constrained by many other cultural forces as well as the Netflix algorithm. Part of the work of the Netflix culture machine is to continually course-correct between that narrow aesthetic littoral and the vast ocean of abstraction behind it, populated by billions of data points and probabilistic inferences that are much more difficult to read, even for system architects like Todd Yellin.

当然，这一论点也受到所有娱乐作品本身也是文化机器这一事实的限制。每个商业文化企业要想成功，都必须做同样的工作：通过参考、回应、借鉴和窃取其他作品和其他受众的经验来聚集观众，并利用类型、道德和悬念等文化规则来吸引观众。解读《纸牌屋》的更大目的并非在算法和电影摄影之间建立因果关系，而是要表明营销和推荐框架本身已经成为一种新的创作形式，就像正在改写华尔街规则的股票交易算法一样（参见第五章）。抽象的美学贯穿了整部剧从构思到播出的整个过程，要求观众具备新的素养，才能参与到Netflix文化机器的工作中。

And of course this argument is also constrained by the fact that all entertainments are also culture machines. Every commercial, cultural enterprise must do the same work to succeed: assemble an audience by referencing, responding to, borrowing, and stealing from other works, other audiences, and by leveraging the cultural rules of genre, morality, and suspense to draw in the crowds. The larger purpose of this exercise in reading House of Cards is not to draw a causal link between algorithms and cinematography, but rather to demonstrate that the marketing and recommendation framework itself has become a new form of authorship, like the stock-trading algorithms that are rewriting the rules of Wall Street (see chapter 5). The aesthetic of abstraction permeates the show from inception to delivery, demanding a new literacy from viewers in order to participate in the work of the Netflix culture machine.

从美学到套利

From Aesthetics to Arbitrage

正如哈贝马斯在《交往行为理论》中所说，当代社会的杠杆点和批判洞察力在于“系统与生活世界的接缝处” ^。47《纸牌屋》的“腐败个性化”只是传统营销在计算媒体作品的屏幕背后广泛消失的一个小例子，这些作品往往模仿更传统的形式和流派。例如，人口和地理集群（例如，15-25 岁的郊区白人男性）正在被新的、快速变化的、大多不透明的名称所取代，这些名称用于制作新型广告。这些类别非常具体，并且故意向我们隐藏：一名男子收到 OfficeMax 发来的一封邮件，地址是“女儿在车祸中丧生”。48^对于普通的 Netflix 用户来说，这也意味着我们不再根据我们自己选择的标准来识别（例如，我们选择在消费者调查中分享的内容），而是根据一系列行为选择来识别，这些行为选择的后果基本上是未知的。你或许会声称自己喜欢浪漫喜剧，但如果你一个夏天看了十遍《黑客帝国》，Netflix 又会觉得你热爱浪漫喜剧有多大价值呢？如果 Netflix 需要推广的新剧是《纸牌屋》 ，Netflix 又会觉得它有多大价值呢？谁来衡量“幸福”？用同样的系统来量化其他一切。

As Habermas argues in The Theory of Communicative Action, the points of leverage and critical insight for contemporary society lie at “the seams between system and lifeworld.”⁴⁷ The “corrupt personalization” of House of Cards is just a small example of the broader disappearance of traditional marketing behind a screen of computational media works that often emulate more traditional forms and genres. For example, demographic and geographical clusters (e.g., white suburban males 15–25) are being replaced by new, rapidly shifting, and mostly opaque designations that are used to produce new kinds of advertising. These categories are incredibly specific and deliberately hidden from us: one man received a mailer from OfficeMax revealingly addressed to him with the suffix “daughter killed in car crash.”⁴⁸ For the average Netflix subscriber, this also means we are no longer identified according to metrics we might choose ourselves (e.g., what we elect to share on a consumer survey) but according to a set of behavioral choices whose consequences are largely unknown. You might claim you love romantic comedies, but how much value does Netflix attribute to that statement if you watch The Matrix ten times in a summer? And how much value does Netflix ascribe to it if the new show Netflix needs to market is House of Cards? Who measures “happiness”? The same systems that quantify everything else.

我们在数字平台上留下的评判、纵容和其他文化决策点的痕迹越来越长，这会导致算法对数字自我进行组合，这些自我既高度具体化，又高度抽象化。大数据允许更精细的抽象，构建消费市场，在这些市场中，我们被定位为个体，或作为临时群体的一部分，这些群体如此具体，以至于掩盖了身份的其他重要方面，例如OfficeMax做出的令人毛骨悚然的决定，向已故儿童的父母推销产品（也许是剪贴簿或其他纪念媒体？）。用营销术语来说，这是市场细分的演变：通过将消费者划分为特定类别，可以独特地定制信息，以回应他们的希望和恐惧。Netflix、谷歌和Facebook等新兴算法平台与更传统的广告之间的区别，从叙事角度来看，是一个难以捉摸的问题。⁴⁹正如漫画理论家斯科特·麦克劳德 (Scott McCloud) 所说，这些系统不仅控制信息本身，还控制信息的框架，通过掩盖干净的界面和简单的选择背后的大部分计算，重塑了哈贝马斯所说的“生活形式的语法”。⁵⁰广告、政治信息甚至金融决策所针对的背景现在可能非常具体，但仍然是看不见的。

The ever-lengthening trail of judgments, indulgences, and other cultural decision points we leave behind on digital platforms can lead to the algorithmic assemblage of digital selves that are both highly specific and highly abstract. Big data allows for more granular abstractions, constructs of consumer markets where we are targeted as individuals, or as part of ad hoc groups that are so specific as to eclipse other important aspects of identity, like OfficeMax’s macabre decision to market products (scrapbooks or other memorial media, perhaps?) to parents of deceased children. In marketing terms, this is the evolution of segmentation: the idea that by clustering consumers into particular categories, messages can be uniquely crafted to address their hopes and fears. The distinction between emerging algorithmic platforms like Netflix, Google, and Facebook and more traditional advertising is in narrative terms a gutter problem.⁴⁹ As comics theorist Scott McCloud might say, these systems control not just the message itself, but the frame around the message, reshaping what Habermas calls the “grammar of forms of life” by obscuring most of the computation behind clean interfaces and simple choices.⁵⁰ The contexts in which we are addressed by advertisements, political messages, and even financial decisions might now be incredibly specific, but remain invisible.

随着我们的文化生活越来越多地在线化，数字平台通过内容的复制和过滤，创造了新的语法结构。Netflix 和 Facebook 在媒体过滤项目上都采用了包容性共创的修辞，识别特定产品或新闻条目如何出现在你的屏幕上（例如，“乔治·史密斯和其他五位朋友点赞了这条信息”）。然而，这些包容性的时刻掩盖了许多其他我们未被邀请参与的决定，尤其是那些为我们的活动划定界限或框架的决定。这些持续不断的邀请放大了杰弗里·鲍克所说的档案“隐形排他性”的逻辑，“它将自己呈现为所有可能陈述的集合，而不是可以表达的规律。” ⁵¹

As we live more of our cultural lives online, digital platforms are making new grammatical structures possible through the replication and filtering of content. Both Netflix and Facebook adopt a rhetoric of inclusive cocreation when it comes to the project of filtering media, identifying how particular products or news items arrive on your screen (e.g., “George Smith and five other friends liked this”). Yet these moments of inclusion mask the many other decisions we are not invited to participate in, particularly those that draw the gutter or frame around our activities. These constant invitations amplify what Geoffrey Bowker calls the “invisibly exclusionary” logic of the archive, “which presents itself as being the set of all possible statements, rather than the law of what can be said.”⁵¹

最成功的算法企业已经将我们一直在探索的计算与文化之间的差距视为机遇，或者我称之为算法套利。随着计算系统变得越来越高效，以及我们身后个人数据的厚重感越来越强，这种套利在我们文化生活中的存在正在迅速扩大，并开始重塑永恒的消费者当下，即“当下”的真正含义。毕竟，围绕着谁能为你构建当下，数十亿美元的交易正在发生。当你访问一个网站，或许是为了了解世界上“此刻”正在发生的事情时，数百台服务器参与了持续几分之一秒的拍卖，以确定哪些广告会出现在页面上，甚至可能根据预测你对不同主题的兴趣的模型来组织页面内容。

The most successful algorithmic businesses have exploited the gap we have been exploring between computation and culture as an opportunity, or what I term algorithmic arbitrage. As computational systems become more efficient, and the patina of personal data we leave behind us grows thicker, the presence of this arbitrage in our cultural lives is rapidly expanding, and beginning to reinvent what the eternal consumer present, the moment of “now,” actually means. After all, there are billions of dollars changing hands over the question of who gets to construct the present for you. When you access a website, perhaps to find out what is happening in the world “right now,” hundreds of servers are involved in auctions lasting fractions of a second to determine which advertisements will appear on the page, and maybe even organize its content according to models predicting your interest in different topics.

算法套利依靠理解差距和文化延迟来获取利润或有价值的信息。“腐败的个性化”以及我们与 Facebook、Google 等公司分享个人数据流的未经审查的交易都依赖于此类套利形式，它们给我们带来有意义的文化数据（《纸牌屋》、精选的亲朋好友新闻），以换取其他信息（我们的兴趣、位置、搜索历史、观看习惯等），而这些信息的价值对我们来说实际上并不清楚，但提供这些服务的公司却知道。当我们采用算法套利所倡导的信息语法时，算法套利会取得最彻底的成功。2011 年，在阿拉伯之春最激烈的时候，Facebook 成为埃及军方的主要媒体机构，该平台再次成为建立公共相关性的文化机器，创造了一种“知识逻辑”，它不仅定义了辩论的术语，还定义了文化表达的元结构。⁵²解放广场抗议活动开始仅三周后，最高指挥部就创建了这个页面，并将其献给“点燃 1 月 25 日革命的埃及子民和青年，以及革命的烈士”。53^军方一定意识到，在努力接触突然变得至关重要的群体——该国心怀不满的青年——时，该平台对于克服传统新闻渠道的文化延迟有多么有效。Facebook 迅速成为军方发布重大声明的第一个渠道，成为现在时的新闻机构，它避开了新闻发布会或国家媒体公告的正式形式，转而通过状态更新、点赞和公开评论等逻辑来发表声明。此举实际上支持了那些将埃及动乱称为“Facebook 革命”的人。

Algorithmic arbitrage depends on gaps of understanding and cultural latency to generate profit or valuable information. “Corrupt personalization” and the unexamined bargains we make to share our personal data streams with companies like Facebook and Google depend on such forms of arbitrage, bringing us meaningful cultural data (House of Cards, curated news about family and friends) in exchange for other information (our interests, locations, search histories, viewing habits, etc.) whose value is effectively unknown to us, but known to the companies providing these services. Algorithmic arbitrage succeeds most completely when we adopt the grammars of information that they espouse. When Facebook became the primary media organ for the Egyptian military during the height of the country’s Arab Spring in 2011, the platform became, once again, a culture machine for establishing public relevance, creating a “knowledge logic” that defined not just the terms of debate but the metastructure of cultural expression.⁵² Just three weeks after protests began in Tahrir Square, the high command created the page and dedicated it “to the sons and youth of Egypt who ignited the January 25 revolution and to its martyrs.”⁵³ The military must have realized how effective the platform was for overcoming the cultural latency of more traditional news channels as it struggled to reach a suddenly crucial demographic, the country’s disaffected youth. Facebook quickly became the military’s first outlet for major announcements, the organ of news in the present tense, eschewing the formality of press conferences or state media announcements for statements framed by the logic of status updates, likes, and public comments. The move effectively endorsed those who called Egypt’s upheaval the “Facebook Revolution.”

这些系统提供了有限的公共治理空间（例如，允许 Facebook 用户通过“喜欢”来推广特定事业），但它们看似民主的界面只是算法套利更深层次大厦的门面。Facebook、谷歌、Netflix 和其他公司通常不会进行公开审查，而是通过算法来策划他们希望我们看到的内容，媒体学者 Ganaele Langlois 称之为“管理意义程度和赋予文化价值”。54^与PageRank 算法和谷歌为防止被谷歌以外的任何人利用而采取的诸多干预措施一样，这些套利系统将用户赋权与严格的信息控制相结合，以鼓励特定行为并隐藏边缘和粗糙之处。这些语法呼应了理论家 Lev Manovich 对新媒体语言的框架，新媒体在多个领域运作：首先，定义了新数字公共领域的共享和协作的数字修辞；其次，计算机科学家和软件工程师隐秘的、祭司般的语言；第三，机器学习算法、海量数据集和随机信息处理系统的计算语言。55在这些语言和试图阅读这些语言的人之间的每一个抽象层上都出现了套利的机会^。

These systems present a limited space of public governance (e.g., allowing Facebook users to promote particular causes through “liking” them), but their seemingly democratic interfaces are facades for the much deeper edifice of algorithmic arbitrage. Facebook, Google, Netflix, and the rest do not often engage in overt censorship, but rather algorithmically curate the content they wish us to see, a process media scholar Ganaele Langlois terms “the management of degrees of meaningfulness and the attribution of cultural value.”⁵⁴ Like the PageRank algorithm and the many interventions Google makes to prevent its exploitation by anyone other than Google, these systems for arbitrage mix user empowerment with strict informational control to encourage particular behaviors and hide the margins and rough edges away. These grammars echo theorist Lev Manovich’s framing of the language of new media, which operates in multiple registers: first, the digital rhetoric of sharing and collaboration that has defined a new digital public sphere; second, the sequestered, priestly language of computer scientists and software engineers; and third, the computational languages of machine learning algorithms, vast data sets, and stochastic information processing systems.⁵⁵ Opportunities for arbitrage emerge at every layer of abstraction between these languages and those attempting to read them.

正如《纸牌屋》在某种意义上是通过算法和人类参与者（包括Netflix的数百万客户）的反馈回路创作的一样，还有许多其他算法系统明确地以劳动和套利的形式吸引人类。它们是文化机器，自诩为人类内容的计算策展人，但实际上却是更为复杂的协作系统，协调着数百万用户、企业或集体目标以及精心抽象的知识结构。维基百科和数字抗议集体“匿名者”都具有这样的功能。这些机器执行的文化工作是操纵不同层次的意义及其之间的实施差距：推荐观看队列、量子理论数据库、投资未来热门节目的高管。如果Netflix让我们瞥见了这些抽象模式的美学，那么更深入地研究算法套利将揭示算法时代的政治经济学，并最终揭示算法影响价值的本质。

Just as House of Cards was, in a sense, created through a feedback loop of algorithmic and human actors that includes Netflix’s millions of customers, there are many other algorithmic systems that explicitly engage humans in forms of labor and arbitrage. They are culture machines that present themselves as computational curators of human content but actually function as much more complex collaborative systems coordinating millions of users, corporate or collective objectives, and elaborately abstracted structures of knowledge. Wikipedia and the digital protest collective Anonymous both function like this. The cultural work such machines perform is to manipulate different layers of meaning and the implementation gaps between them: the recommended viewing queue, the Quantum Theory database, the executives investing in a future hit show. If Netflix has allowed us to glimpse the aesthetics of these modes of abstraction, a deeper look at algorithmic arbitrage will reveal the political economy of the algorithmic era and, ultimately, the nature of algorithmically inflected value.

笔记

Notes

感觉就像一句俏皮话

It Felt Like a One-Liner

2007年，我在斯坦福大学攻读研究生学位时，几乎从未想过，就在不久的将来，本科生和助教们就能通过为一个名为Facebook的新平台设计简单的应用程序，一夜之间赚取数千美元。这些学生利用Facebook的应用程序接口（API），编写一个程序来分享诸如拥抱或“热点”之类的内容，通常只需几个小时就能完成。2^当他们开始在应用程序中投放广告时，资金便源源不断地涌入，在一些教室和宿舍里营造出一种“淘金热”的氛围^。3

When I was pursuing graduate studies at Stanford University in 2007, I had little idea that just around the corner undergraduates and teaching assistants were making thousands of dollars overnight by designing simple applications for a new platform called Facebook. Using the company’s application program interface (API), these students would code a program to share things like hugs or “hotness points,” often in a matter of hours.² Once they began placing ads in the applications, money flooded, contributing to a “gold rush” atmosphere in a few classrooms and dorms.³

但是，是什么促使所有这些用户互相发送热点，参与到这些公共关注网络中呢？这些课程的负责人之一是 BJ Fogg，他是一位著名的网络行为心理学家，负责斯坦福大学的说服技术实验室。Fogg 在“捕获学”和说服技术方面的研究表明，像 Facebook 这样的工具之所以能够产生深远的行为影响，是因为它们能够触及重要的心理触发因素。他将“说服性设计”定义为个人动机的支点、行动的能力或潜力，以及能够促使其采取特定行动的特定触发因素。4开发这些早期应用程序的^学生创造出的产品吸引了成千上万的人参与到对广告商有利可图的特定形式的工作中：点击按钮、与好友分享链接或其他鼓励参与的简单机制。这些应用程序通常请求访问用户的 Facebook 好友网络，从而允许软件向其他联系人发送消息，并利用每个用户的关系来扩大其覆盖范围。其中许多游戏对玩家的价值有限，但对开发者来说却像是一种有利可图的文化病毒，可以在社交网络上迅速传播定制广告。

But what compelled all those users to send one another hotness points, to participate in these public networks of attention? One person leading these classes was B. J. Fogg, a noted psychologist of online behavior who runs the Persuasive Technology Lab at Stanford. Fogg’s work on “captology” and persuasive technology argues that tools like Facebook have profound behavioral effects because they tap into significant mental triggers. He identifies “persuasive design” as the fulcrum of a person’s motivation, her ability or capacity to act and the specific triggers that could push her into taking particular actions.⁴ Students building these early apps created products that induced thousands of people to engage in particular forms of work that are lucrative to advertisers: clicking on a button, sharing a link with friends, or another simple mechanic to encourage engagement. These apps typically request access to a user’s network of Facebook friends, allowing the software to send messages to other contacts and leverage each user’s relationships to extend its reach. Many of them offer limited value to their players but function as lucrative cultural viruses for their developers, spreading tailored advertising rapidly across social networks.

自早期以来，“社交游戏”已变得越来越复杂，利润也越来越丰厚，吸引了众多参与者的关注，从大型电子游戏工作室到社会活动家，他们都希望在数百万人中催化特定形式的文化工作。随着社交游戏本身逐渐成为一个重要的市场，游戏的界限也变得越来越模糊。这种转变通常被定义为“游戏化”——利用受游戏玩法启发的因果机制来鼓励特定行为。正如一位领先的倡导者所定义的那样，

Since those early days, “social gaming” has become more sophisticated and more profitable, drawing the attention of a huge range of actors, from major videogame studios to social activists, all hoping to catalyze particular forms of cultural work among millions of people. As social gaming has risen to become a significant market in its own right, the boundary lines of play continue to blur. This transition has often been framed through the term “gamification”—the idea of using cause and effect mechanisms inspired by game-play to encourage particular behaviors. As one leading advocate defines it,

游戏化可以理解为利用游戏系统的某些元素来实现商业目标。最容易识别这一趋势的体验（例如常旅客计划、Nike Running/Nike+ 或 Foursquare）能够立即带来游戏般的体验。积分、徽章、等级、挑战、排行榜、奖励和新手引导等关键游戏机制的存在，都预示着游戏正在进行。然而，游戏化正越来越多地被用于创造利用游戏力量但又不那么直白的体验^。5

gamification can be thought of as using some elements of game systems in the cause of a business objective. It’s easiest to identify the trend with experiences (frequent flyer programs, Nike Running/Nike+ or Foursquare) that feel immediately game-like. The presence of key game mechanics, such as points, badges, levels, challenges, leader boards, rewards and onboarding, are signals that a game is taking place. Increasingly however, gamification is being used to create experiences that use the power of games without being quite as explicit.⁵

再次借用哈贝马斯的术语，游戏化是将系统刻意嫁接到生活世界，创造出一个由指标和任意目标组成的上层建筑，这些指标和目标与文化行为紧密相连。Facebook 上充满了这些线索，有些线索比其他线索更微妙，它们将友谊和沟通的抽象概念转化为细粒度、可量化的行动。从积累的好友数量到宣布与特定粉丝群体或政治事业的联系，Facebook 是一个庞大的参与度衡量系统。在帖子或照片中标记其他用户成为一种社交诱导，给予对方精神上的轻微提升，并带来被公众认可的愉悦感^。6

Gamification is a deliberate grafting of system onto lifeworld, using Habermas’s terms again, creating a superstructure of metrics and arbitrary goals attached to cultural behaviors. Facebook is full of these cues, some more subtle than others, for translating abstract conceptions of friendship and communication into granular, countable actions. From the number of friends one accumulates to declaring affiliations with particular fan groups or political causes, Facebook is a vast system for measuring engagement. Tagging another user in a post or a photograph becomes a kind of social grooming, giving that person a small mental boost and the pleasure of being publicly recognized.⁶

但Facebook上的游戏化在数百万用户使用的第三方应用中达到了顶峰。或许最著名的例子是Zynga于2009年发布的令人上瘾的社交农场游戏《FarmVille 》，巅峰时期吸引了8000万玩家^。7作为游戏化的经典案例，《FarmVille》通过设置一系列提示和奖励来吸引玩家长时间的持续参与。游戏规则将某些操作推迟，要求玩家在数小时或数天后回来收割庄稼或进行其他维护活动。这种对典型在线互动异步灵活性的扭曲，本身就对玩家产生了诱惑，导致一些玩家半夜醒来照料他们的虚拟农场。对于一小部分玩家来说，解决方案是支付小额费用，至少暂时免除这些苛刻的规则，例如让庄稼立即生长。这一小部分玩家支撑了整个商业模式，每年创造了数亿美元的收入。FarmVille及其后续作品能够有效地通过胡萝卜加大棒的结合，引发人类特定的固定行为，让玩家参与到公司可以直接赚钱或用来扩大用户网络的活动中。

But gamification on Facebook reaches its apogee in the third-party applications that millions of people use on the site. Perhaps the most notorious example is FarmVille, the addictive social farming game released by Zynga in 2009, which at its peak attracted 80 million players.⁷ A classic example of gamification, FarmVille hooked players by creating a set of cues and rewards for sustained engagement over long periods of time. Certain actions were deferred by the game rules, requiring a player to come back hours or days later to harvest their crops or perform other maintenance activities. This perversion of the asynchronous flexibility of typical online interaction created its own seductive rigor for players, leading some to wake up in the middle of the night to tend to their virtual farms. For a certain minority of players, the solution was to make small payments that would waive these oppressive rules, at least temporarily, allowing a crop to grow instantly, for example. That minority has supported the entire business model, generating hundreds of millions of dollars in annual revenue. FarmVille and its successors are effective at eliciting particular rote behaviors from humans through a combination of carrots and sticks, engaging them in actions that the company can monetize directly or use to expand its network of users.

农场的文化叙事层面掩盖了一个令人着迷的斯金纳箱——经典的操作性条件反射工具，在本例中，它将创收行为与人类与生俱来的社会联系和完成感联系起来。正如《 大西洋月刊》援引一位公司高管的话：“玩Zynga游戏最引人入胜的部分之一，就是决定何时以及如何提醒好友玩Zynga游戏。” ⁸归根结底，这些游戏是一种伪装成效率的逃避现实——种植庄稼，建立帝国，在排队等候的60秒内完成所有任务。但这种颗粒状、渐进式的逃避现实掩盖了其自身的纪律性和生产力。 《 FarmVille》既是Facebook的一部分，又超越了Facebook，利用该网站持续不断的用户互动模式，吸引了大量用户的注意力和金钱。在极端情况下，它们甚至会滋生出一些成瘾行为，其程度堪比毒品和酒精对个人的毁灭^。9

The cultural narrative layer of the farm masks a mesmerizing Skinner box, the classic tool of operant conditioning, which, in this case, links revenue-generating behaviors to the innate human rewards of social connectedness and completion. As The Atlantic paraphrased one company executive: “One of the most compelling parts of playing Zynga’s games is deciding when and how to spam your friend with reminders to play Zynga’s games.”⁸ Ultimately these games are a kind of escapism masquerading as efficiency—plant your crops, build your empire, complete this task all in sixty seconds while you wait in line. But that pelletized, incremental escapism obscures its own forms of discipline and productivity. FarmVille is both of and beyond Facebook, leveraging the site’s persistent modes of contact with users to capture a shocking amount of their attention and cash. In the extreme, they can foster forms of addiction that approach the personal destruction of drugs and alcohol.⁹

伊恩·博格斯特（Ian Bogost）一直对游戏化持批评态度，他认为游戏化应该被重新定义为“剥削软件”（exploitationware），因为它滥用了人类易受玩世不恭的营销人员操纵的本性。但与其他人一样，他也承认游戏具有激励行为的潜力：“即使是对电子游戏的谴责也承认它们蕴含着特殊的力量，这种力量能够吸引我们，让我们沉迷其中，鼓励我们重复似乎已经做过的事情，让我们为看似不存在的东西花钱。” ¹⁰ “剥削软件”一词​​之所以有用，是因为它突显了这些游戏本质上的商业性，表明它们作为算法文化机器的角色，能够通过与用户的互动有效地挖掘或提取特定形式的价值。

Ian Bogost has been a persistent critic of gamification, arguing that it should be reframed as “exploitationware” for its abuse of human susceptibility to manipulation by cynical marketers. But like others, he acknowledges the potency of games to motivate behaviors: “even condemnations of video games acknowledge that they contain special power, power to captivate us and draw us in, power to encourage us to repeat things we’ve seemingly done before, power to get us to spend money on things that seem not to exist.”¹⁰ The term exploitationware is useful because it highlights the essentially commercial aspect of these games, signaling their role as algorithmic culture machines that effectively mine or extract particular forms of value through interactions with users.

正是对这种趋势的回应，博格斯特创造了自己的一款同样“具有说服力”的开发软件。“奶牛点击器”最初是对这些社交游戏无意义重复的讽刺：

It was in reaction to this trend that Bogost created his own piece of exploitationware that was also “persuasive.” Cow Clicker began as a satirical response to the mindless repetition of these social games:

像《FarmVille》这样的游戏就是奶牛点击游戏。你只需要点击一头牛，就这么简单。我记得当时觉得这就像一句俏皮话，就像那种你会在推特上说的。我就把它记在了脑子里^。11

Games like FarmVille are cow clickers. You click on a cow, and that’s all you do. I remember thinking at the time that it felt like a one-liner, the kind of thing you would tweet. I just put it in the back of my mind.¹¹

但随着博格斯特对Zynga及其同类产品的公开批评引起更多关注，他决定用一款模仿操纵性社交游戏最糟糕一面的游戏来展示自己的想法。正如记者杰森·坦兹在《连线》杂志上描述的整个过程：

But as Bogost’s public critique of Zynga and its ilk drew more attention, he decided to demonstrate his ideas with a game that emulated the worst excesses of manipulative social games. As journalist Jason Tanz described the whole adventure for Wired:

规则简单得近乎荒谬：游戏中有一张奶牛的图片，玩家每六个小时可以点击一次。每点击一次，玩家就能获得一分，称为一次点击。玩家最多可以邀请八位好友加入他们的“牧场”；牧场里任何人点击他们的奶牛，都能获得一次点击。排行榜追踪游戏中点击次数最多的玩家。玩家可以购买游戏币，称为“mooney”，用来购买更多奶牛或规避时间限制。游戏秉承了《FarmVille》的一贯风格，每当玩家点击一头奶牛时，他们的Facebook动态消息中就会出现一条公告——“我正在点击一头奶牛” ^。12

The rules were simple to the point of absurdity: There was a picture of a cow, which players were allowed to click once every six hours. Each time they did, they received one point, called a click. Players could invite as many as eight friends to join their “pasture”; whenever anyone within the pasture clicked their cow, they all received a click. A leaderboard tracked the game’s most prodigious clickers. Players could purchase in-game currency, called mooney, which they could use to buy more cows or circumvent the time restriction. In true FarmVille fashion, whenever a player clicked a cow, an announcement—“I’m clicking a cow”—appeared on their Facebook newsfeed.¹²

《奶牛点击器》被刻意设计得荒诞不经，是一款揭露游戏化虚伪与操控的毫无意义的游戏。但它却火了，起初是作为与博格斯特观点相同的人的讽刺性抗议，后来它成为了一款独立的游戏。玩家要么没有意识到这是讽刺，要么明知故犯地玩，就像那位居家父亲告诉坦兹的那样：“与其玩毫无意义的愚蠢游戏，不如玩有意义的愚蠢游戏。” ¹³在它的鼎盛时期，超过 5 万人点击数字奶牛，博格斯特发现自己陷入了自己建立的斯金纳箱式反馈系统，每当他为游戏添加新功能时，都会得到玩家社区的奖励。博格斯特将这个过程描述为一种类似于方法表演的“方法设计”，将自己置于社交游戏设计师的创作空间中，并最终遭受与软件相同的系统性、非人性化的纠缠，他看到软件给玩家带来了同样的后果：“我很难表达出在创作这个看似微不足道的小理论兼戏仿游戏时，伴随它的是怎样的冲动和自我厌恶。” ¹⁴

Cow Clicker was deliberately designed to be absurd, a meaningless game that would reveal the hypocrisy and manipulation of gamification. But it became popular, first as an ironic protest from others who shared Bogost’s views, then as a game in its own right. Players either didn’t realize that this was a satire, or played in spite of that knowledge, like the stay-at-home father who told Tanz, “instead of stupid games that have no point, we might as well play a stupid game that has a point.”¹³ At its apogee, over 50,000 people were clicking on digital cows and Bogost found himself enmeshed in his own Skinner box of feedback, getting rewarded by the player community when he added new features to the game. Bogost has described this process as a kind of “method design” like method acting, putting himself into the creative space of a social game designer and ultimately suffering the same kind of systemic, dehumanizing entanglement with the software that he sees it inflicting on players: “It’s hard for me to express the compulsion and self-loathing that have accompanied the apparently trivial creation of this little theory-cum-parody game.”¹⁴

对博格斯特来说，对社交游戏现象及其所依赖的社交网络的批判，其核心在于技术哲学的另一个基石——马丁·海德格尔的“框架”概念。简而言之，海德格尔认为，技术（以及我们整个社会世界）往往会促使我们进入某种思维模式，思考宇宙中什么是可能的，以及什么可以被揭示。我们看到一把锤子，就会思考它能用来敲什么；但锤子也可以用来开瓶子、撑门，或者在刮风的日子里压住纸张。

For Bogost, the heart of this critique of the social gaming phenomenon and the social networks it relies on is another foundation-stone in the philosophy of technology, Martin Heidegger’s notion of enframing. In very simple terms, Heidegger argued that technologies (and our social world in general) tend to nudge us into certain modes of thinking about what is possible and what can be revealed about the universe. We see a hammer and we think about what we can hammer with it; but a hammer could also be used to open a bottle, to prop open a door, to hold down papers on a windy day.

在《单元操作》一书中，博格斯特描述了社交网络如何通过明确设计社交关系的表征及其操纵工具来鼓励框架模式：“在领英上，将一个商业伙伴介绍给另一个商业伙伴突然变成了一个正式的单元操作：一系列软件交互，在巩固用户个人体验的同时，也促成了更广阔的职业网络。” ¹⁵ 《奶牛点击器》将这种框架模式推向了一个逻辑上荒谬的结局，玩家可以通过好友“点击你的点击”来积累额外积分，从而将点击奶牛这一毫无意义的行为形式化。用博格斯特的话来说，单元操作几乎抽空了所有真实内容，只留下了讽刺的数字奶牛，而网络基础设施，即一款令人上瘾的社交游戏的程序操作，仍然高度可见。博格斯特实验失控的方式反映了海德格尔和西蒙东等哲学家发起的关于技术性的持续争论，但现在这些争论在状态更新和社交游戏中得到了充分体现。随着用户努力应对框架模式带来的智力和情感后果，一场意义之战由此开启。Cow Clicker失控的符号学表明，这是讽刺和真诚的追求，是游戏和古拉格的结合。

In Unit Operations, Bogost describes how social networks encourage modes of enframing by explicitly designing representations of social relationships and tools for manipulating them: “In LinkedIn introducing one business associate to another suddenly becomes a formal unit operation: a set of software interactions that enable bigger professional networks while fixing users’ individual experiences.”¹⁵ Cow Clicker takes this enframing to a logically absurd conclusion, allowing players to accumulate extra points when friends “click on your clicks,” formalizing the meaningless action of clicking on a cow. In Bogost’s terms, the unit operation is evacuated of almost all real content, leaving only the satirical digital cow behind, while the network infrastructure, the procedural operations of an addictive social game, remain highly visible. The way that Bogost’s experiment spiraled out of control reflects the ongoing debate about technicity launched by philosophers like Heidegger and Simondon, but now played out in status updates and social games. As users grapple with the intellectual and emotional consequences of enframing, a battle for meaning emerges. The runaway semiotics of Cow Clicker signal that this is satire and sincere pursuit, game and gulag all at once.

博格斯特对这些框架模式的批判，探讨了社交参与、个人冲动和交互设计之间令人不安的交集。《Cow Clicker》在多个层面上都是一款糟糕的游戏：它的设计故意糟糕且无趣；它明确地旨在通过任意设定的六小时截止时间浪费玩家的时间；它无耻地利用玩家的社交网络来扩大其病毒式传播范围；它诱使玩家为讽刺性的“mooney”花钱，以至于一位评论者指出：“让我着迷的是，你从中赚的钱越多，你就会越沮丧。我喜欢这一点，我觉得这很有趣。” ¹⁶但《Cow Clicker》不仅仅是一款游戏——它讽刺了一种更深层次的不断发展的文化关系，这种关系随着智能手机和普适计算时代的到来而蓬勃发展，成为一种新的边缘系统性殖民。与《FarmVille》一样，《Cow Clicker》呼吁人们关注一系列文化交易，这些交易模糊了时间、文化和财务价值单位之间的界限，创建了算法套利系统，从“游戏”空间中获取注意力和收入。正如媒体学者麦肯齐·沃克对 Facebook 核心结构的争议性描述：“向量阶层（在本例中指 Zynga 和 Facebook）的权力不再直接拥有文化产品，而是围绕对向量的控制进行巩固。我们拥有所有的文化，他们获得所有的收入。” ¹⁷我们可以玩《FarmVille》，创建自己独特的虚拟家园，但最终要为这种特权付出时间、社会地位（通过与 Zynga 分享我们的网络）以及金钱的代价。

Bogost’s critique of these modes of enframing explores the troubling intersection of social engagement, personal compulsion, and interactive design. Cow Clicker is a bad game on multiple levels: its design is deliberately poor and uninteresting; it explicitly aims to waste its players’ time through arbitrary six-hour deadlines; it shamelessly leverages their social networks to expand its viral reach; it tempts them into spending real money on satirical “mooney,” leading one commenter to note: “What fascinates me is the fact that the more money you make from this, the more depressed you are going to feel. I like that, I think it’s funny.”¹⁶ But Cow Clicker is not just a game—it satirizes a much deeper form of evolving cultural relationship, one that has blossomed with the age of smartphones and ubiquitous computing into a new kind of systemic colonization of the margins. Like FarmVille, Cow Clicker calls attention to a series of cultural transactions that blur the distinctions between temporal, cultural, and financial units of value, creating systems of algorithmic arbitrage that extract attention and revenue from spaces of “play.” As media scholar McKenzie Wark polemically describes the core arrangement on Facebook: “The power of the vectoral class [in this case, Zynga and Facebook] retreats from direct ownership of the cultural product but consolidates around the control of the vector. We get all the culture; they get all the revenue.”¹⁷ We get to play FarmVille, creating our own distinctive virtual homesteads, but end up paying for the privilege with time, social status (by sharing our networks with Zynga), and often money.

《奶牛点击器》是对社交游戏的批判，但它也揭示了算法如何重构各种职业和社交领域的规则。Facebook、LinkedIn 及相关平台将社交互动提炼成一系列结构清晰的游戏，并通过追踪行为、互动和其他形式反馈的算法架构来统计得分。社交影响力、职业人脉和友谊网络都早于互联网出现，但这些联系的意义、它们如何被计算和解读，如今正被算法平台积极地塑造和定义。Facebook 或许没有明确定义“朋友”是什么，但它会告诉你有多少朋友，并为你推荐新朋友，通过其极其成功的平台，创造了一个强大的、有时甚至令人上瘾的隐性定义。

Cow Clicker is a critique of social games, but it also reveals how algorithms are restructuring rules in all sorts of professional and social arenas. Facebook, LinkedIn, and related platforms distill social interactions into a set of explicitly structured games, and the scores are tallied by algorithmic architectures tracking behaviors, interactions, and other forms of feedback. Social influence, professional networking, and friendship networks all predate the Internet, but the meaning of these connections, the ways that they count and are made legible, now get aggressively shaped and defined by algorithmic platforms. Facebook may not explicitly define what a “friend” is, but it will tell you how many you have and suggest new friends for you, creating a powerful, sometimes even addictive implicit definition through its hugely successful platform.

作为一台文化机器，博格斯特的创作体现了当代注意力市场中劳动与价值观念的冲突。对于Zynga和许多其他算法娱乐公司而言，计算是将人类注意力转化为收入的手段。对于这些游戏的玩家来说，奖励是参与某种数字护理，照料一个虚拟的被照料对象（农场、数字宠物等等），并在实践社群中与其他玩家互动。用户在算法框架的约束下构建自己的叙事，即使他们点击为他们设置的斯金纳箱。算法构建并追踪这些行为，像雨滴一样将它们收集起来，然后作为批量数据商品转售。与此同时，玩家通常只感知到这个更大市场中的一小部分，他们通常不仅贡献自己的注意力（用于观看广告）和社交图谱（用于深化他们在数据经纪人中的个人资料并扩大算法的覆盖范围），还贡献自己的现金，进行游戏内购买以提升游戏体验。对于我们许多人来说——2015 年全球 32 亿互联网用户中，约有 15 亿人每月至少访问一次 Facebook——某种形式的此类交易构成了“乐趣”的主要来源^。18

As a culture machine, Bogost’s creation illustrates the conflicting notions of labor and value in the contemporary marketplace of attention. For Zynga and many other algorithmic entertainment companies, computation is the means for converting human attention into income. For the players of these games, the rewards are to engage in a kind of digital grooming, tending an imaginary object of care (a farm, a digital pet, etc.) and engaging with other players in a community of practice. Users construct their own narratives within the constraints of algorithmic enframing even as they click through the Skinner boxes set up for them. The algorithms structure and track these actions, gathering them like drops of rain in a catchment to be resold as a bulk data commodity. Meanwhile, the players generally perceive only a fragment of this larger market situation, often donating not just their attention (to view ads) and their social graph (to deepen their profiles with data brokers and to expand the algorithm’s reach), but also their cash, making in-game purchases to enhance their playing experience. For many of us—roughly 1.5 billion people accessed Facebook at least once a month in 2015, out of 3.2 billion Internet users worldwide—some version of these transactions constitutes a major source of “fun.”¹⁸

工作与娱乐

Work and Play

作为一件概念艺术作品， 《Cow Clicker》最引人注目的元素或许在于它对“乐趣”概念的颠覆。博格斯特刻意将游戏设计得笨拙乏味，将游戏行为集中在一个几乎完全静止的物体——奶牛身上，它每六个小时只能被点击一次（且不会产生任何直接效果）——这仿佛是数字时代的一种苦行或忏悔仪式。这种对乐趣的抽象化，刻意试图消除任何真正的乐趣，将“游戏”变成了一个极其刻板的斯金纳箱，充斥着肤浅的动作和更肤浅的奖励。但《Cow Clicker》仅仅强调了一种日益盛行的新型游戏性劳动。游戏术语“刷任务”（grinding）指的是在游戏中为了积累资源或获得力量而进行的重复性行为：例如，在像《魔兽世界》这样的游戏中，反复收集相同的物品或完成相同的小挑战。刷任务活动本身通常很无趣，但玩家参与其中是为了解锁之后做更多有趣事情的能力。我们现在做一些无聊的事情是为了以后做一些有趣的事情……这是工作的一种定义，这种定义日益渗透到娱乐领域，并困扰着工作和娱乐之间的分离^。19

Perhaps the most compelling element of Cow Clicker as a work of conceptual art is its inversion of the concept of fun. Bogost designed the game to be deliberately awkward and tedious, centering the action on an almost entirely static object, the cow, that can be clicked (to no direct effect) only once every six hours—a kind of ascetic or penitent ritual of the digital age. This is an abstraction of fun that deliberately seeks to eliminate any real joy, turning the “game” into an especially stark Skinner box of shallow action and even shallower rewards. But Cow Clicker merely accentuates a new kind of ludic labor that has become increasingly prevalent. The gaming term “grinding” describes the performance of repetitive actions to accumulate resources or gain powers within a game: repeatedly gathering the same item or completing the same minor challenge in a game like World of Warcraft, for example. The grinding activities are typically uninteresting in themselves, but players engage in them in order to unlock the ability to do more interesting things later. Boring things we do now in order to do fun things later … this is one definition of work, one that has increasingly penetrated the space of entertainment and troubled the separation of work and play.¹⁹

游戏和正常生活的界限正在变得模糊：玩家在虚拟世界中举办婚礼和葬礼；一些招聘主管将游戏团队管理视为一种领导经验；公司采用积分、徽章和其他游戏化方法来鼓励特定的员工行为。20^这种转变部分源于人口统计学因素：游戏逐渐从青少年的消遣演变为许多成年人毫不掩饰地承认为爱好或痴迷的活动。一份行业报告显示，到 2014 年，游戏玩家的平均年龄为 31 岁。21主张“^我们还不如玩一个有意义的愚蠢游戏”的Cow Clicker玩家巧妙地诠释了游戏研究员兼经济学家 Edward Castronova 所描述的围绕着合成世界的幻想“膜”，这个“近乎魔法的圈子”似乎将游戏空间与常规活动隔离开来，但却从未真正发挥作用。²²这是在计算与文化之间进行的游玩，一个受到相互竞争的政治、文化和计算隐喻影响的空间：广受欢迎的《魔兽世界》可以同时在流行病学、种族研究、组织管理、经济学，当然，也可以作为一种奇幻叙事的框架来解读。正如媒体理论家亚历山大·加洛韦在《游戏：算法文化论文集》中所说，“电子游戏将社会现实转化为可玩的形式。” ²³在这里，计算的魔力触及了社会建构的游戏空间的深层根源，例如嘉年华或剧院。

The boundaries between games and normal life are blurring: players hold weddings and funerals in virtual worlds; some hiring executives consider gaming team management as a form of leadership experience; companies employ points, badges, and other gamification methods to encourage particular kinds of employee behavior.²⁰ Part of the shift is demographic: games have gradually evolved from an adolescent pastime to an activity that many adults unabashedly acknowledge as a hobby or obsession. According to an industry report, by 2014 the average gamer was thirty-one years old.²¹ The Cow Clicker player who argues “we might as well play a stupid game that has a point” neatly illustrates what game researcher and economist Edward Castronova describes as the “membrane” of fantasy surrounding synthetic worlds, the “almost magic circle” that seems to protect the space of play from regular activity but never quite succeeds.²² This is play in the gap between computation and culture, a space inflected by competing political, cultural, and computational metaphors: the ever-popular World of Warcraft can be read at once in the frames of epidemiology, race studies, organizational management, economics, and, of course, as a fantasy narrative. As media theorist Alexander Galloway puts it in Gaming: Essays on Algorithmic Culture, “video games render social realities into playable form.”²³ Here the magic of computation taps the deep roots of socially constructed spaces of play, like the carnival or the theater.

进入这层可渗透的幻想膜，我们便进入了一个空间，在这里，相互交融的规则、信仰和价值观为文化套利提供了机会。例如，正如人类学家邦妮·纳尔迪 (Bonnie Nardi) 所报告的，《魔兽世界》包含众多公会：“基督教公会、同性恋公会、本地公会、家族公会、军事公会、同事公会和专业同事公会”，以及为喜欢更具历史体验的玩家预留的特殊服务器，他们使用“仿古英语方言”进行交流。24^在达到临界质量和集体参与的某个阶段，这个不完整循环的魔力会变得更具炼金术，它利用现实空间和虚拟空间之间的难题，机会主义地将某些形式的价值转化为其他形式。这种炼金术的一个例子就是卡斯特罗诺瓦所说的“社会认同”，即游戏幻想世界中抢手的物品仅仅因为有足够多的人渴望它而获得真正价值的过程。²⁵当足够多的玩家想要游戏中可以转让的某种武器或物品时，eBay 等网站上就会出现一个以“真实”货币交易该物品的市场。26^工作和娱乐的世界相互碰撞，在计算和文化之间创造了新的套利形式：例如，在《魔兽世界》和类似游戏中，所谓的“打金农民”在血汗工厂的条件下“玩游戏”，获取可以卖钱换取虚拟商品。

Entering this permeable membrane of fantasy, we gain access to a space where cross-pollinating rules, beliefs, and values provide their own opportunities for cultural arbitrage. World of Warcraft, for example, contains multitudes, as anthropologist Bonnie Nardi reports: “Christian guilds, gay guilds, location-based guilds, family guilds, military guilds, guilds of coworkers, and guilds of professional colleagues,” as well as special servers set aside for players who prefer a more historical experience, communicating in an “ersatz Ye Olde English patois.”²⁴ At a certain stage of critical mass and collective engagement, the magic of this incomplete circle becomes more alchemical, using the aporias between real and virtual spaces to opportunistically convert certain forms of value into others. One example of this alchemy is what Castronova terms “social validation,” the process by which a sought-after item within the fantasy world of a game acquires real value simply because enough people desire it.²⁵ When enough players desire a particular weapon or artifact that can be transferred in a game, a market emerges on sites like eBay for trading that item in “real” currency.²⁶ The worlds of work and play collide, creating new forms of arbitrage between computation and culture: so-called “gold farmers” in World of Warcraft and similar games, for example, “play” in sweatshop conditions to acquire virtual goods that can be sold for real money.

然而，不同价值观之间炼金般的套利，其对文化空间的渗透远比这些简单的交易更为深刻。虚拟游戏看似随意的规则，却能对受其影响的生活产生深远的影响，改写日常生活的实践。互联网上充斥着网络成瘾的故事：游戏爱好者们在《魔兽世界》或《农场小镇》等虚拟世界的诱惑下，失去了工作、婚姻、积蓄，甚至生命。这些破坏性的成瘾和对算法娱乐的依赖相对罕见，但它们展现了算法系统重塑人类生活、剥夺人类传统意义和归属感的力量。面对引​​人入胜的计算系统，尤其是游戏以及那些带有斯金纳箱式游戏特征的系统，我们常常会感到一种参与其中的冲动。有时，我们几乎没有选择，因为算法系统既承担着工作，也承担着娱乐，例如电话服务台、劳动力调度系统和许多其他应用程序的客户服务和一线管理角色。

The alchemical arbitrage of different values can pierce cultural space much more deeply than these straightforward transactions, however. The seemingly arbitrary rules of virtual games can have a deep impact on the lives touched by that alchemy, rewriting the practices of everyday life. The Internet is filled with stories of Internet addiction: game aficionados who lost their jobs, their marriages, their savings, and even their lives to the call of synthetic worlds like World of Warcraft or FarmVille. These instances of destructive addiction and dependence on algorithmic entertainment are relatively rare, but they demonstrate the power of algorithmic systems to reorder human lives, to evacuate them of traditional forms of meaning and belonging. Faced with appealing computational systems, especially games and those that take on the Skinner box trappings of games, we often feel the compulsion to engage. Sometimes there is little choice involved, as algorithmic systems take on work as well as play via customer service and frontline management roles for telephone helpdesks, workforce scheduling systems, and many other applications.

我们看到，人类正在以所有这些方式与真正的算法空间作斗争。这些系统由计算结构化规则排序，然后通过人类意图和文化重构来操纵这些规则（有时会被黑客入侵或违反，就像 Hiro 在 Metaverse 中挥舞武士刀一样）。在《新媒体语言》中，Lev Manovich 将玩家通过游戏学习算法描述为将计算“转码”为人类行为：“将计算机本体投射到文化本身。” ²⁷但这种体验超越了游戏——它也是一种调查、解释和阅读的行为。在《游戏》中，加洛韦将这种关系作为一种解释学、一种解释知识的模型进行探索。加洛韦的书转向了“解读游戏意味着解读其算法”的说法，从而将游戏的本质意义与其作为计算文化机器的地位联系起来。²⁸玩家在掌握《侠盗猎车手 V》的规则时，就像评论家撰写文章一样，也在解释游戏的算法——他们都在进行一种既是工作又是娱乐的解释行为，而这种劳动的一部分是跨越计算和文化意义系统之间的差距所需的努力。

In all of these ways we see humans grappling with truly algorithmic spaces. These systems are ordered by computationally structured rules that are then manipulated (and at times hacked or contravened, like Hiro wielding his katana in the Metaverse) by human intention and cultural reframing. In The Language of New Media, Lev Manovich describes a player learning an algorithm through play as a kind of “transcoding” of computation into human behavior: “the projection of the ontology of a computer onto culture itself.”²⁷ But this experience moves beyond play—it is also an act of investigation, of interpretation, of reading. Galloway explores this relationship as a kind of hermeneutics, a model for interpreting knowledge, in Gaming. Galloway’s book turns on the claim that “to interpret a game means to interpret its algorithm,” thus linking the essential meaning of a game to its status as a computational culture machine.²⁸ The player coming to grips with the rules of Grand Theft Auto V is interpreting the game’s algorithm as much as the critic writing about it—they both perform a hermeneutic act that is both work and play, and part of that labor is the effort required to span the gap between computational and cultural systems of meaning.

加洛韦将工作与娱乐之间的这种紧张关系置于“信息控制”的背景下构建，玩家要么在《文明》等游戏中扮演角色（管理数百万数字公民的生活），要么在《奶牛点击器》等游戏中屈服于信息控制（等待⁶小时的计时器滴答作响）。29^这种紧张关系可以追溯到诺伯特·维纳和控制论：在他的职业生涯中，维纳越来越关注控制论在实施中的后果，特别是在自动化和劳动力方面。30控制论中反馈回路和有机体作为信息实体的理想也可以应用于摩尼教系统，即操纵人类参与者以达到令人讨厌或仅仅是不诚实的目的。科幻小说作家道格拉斯·亚当斯非常喜欢根据《银河系漫游指南》改编的文字冒险游戏，因为它已经从“用户友好”变成了“侮辱用户”和“用户撒谎”。31^玩家觉得它很有趣。

Galloway frames this tension between work and play in the context of “informatic control,” which players either role play in games like Civilization (managing the lives of millions of digital citizens) or submit to in games like Cow Clicker (waiting for that six-hour counter to tick down).²⁹ This tension reaches back to Norbert Wiener and cybernetics: over the course of his career, Wiener grew increasingly concerned with the consequences of cybernetics in implementation, particularly around automation and labor.³⁰ The cybernetic ideal of the feedback loop and the organism as an informational entity could also be applied to Manichean systems that manipulate human participants for unsavory or merely dishonest ends. The science fiction author Douglas Adams took great pleasure in a text adventure game based on The Hitchhiker’s Guide to the Galaxy because it had moved beyond “user friendly” into “user insulting” and “user mendacious.”³¹ Players found it delightful.

正如加洛韦等人所言，对游戏化的真正批判源于这种殖民进程的逻辑延伸，因为游戏化不仅定义了社交互动，还定义了劳动和社会的深层结构。游戏最引人注目的方面之一，恰恰在于算法排序宇宙的诱惑——在这些空间里，我们可以尽情沉迷于幻想性错觉，所有事件和过程都按照一套规则运行。这些宇宙在美学上井然有序，其中的规则和条件，我们相信是可以学习并最终掌握的。计算秩序的美学呼应了博格斯特关于大教堂的警告；它对人类参与的吸引力正迅速从娱乐扩展到工作。越来越多的初创公司将这种逻辑带入现实世界，为出租车（例如Uber和Lyft）、家务（Handy、HomeJoy、Mopp）甚至办公室通讯（Slack）等服务创造类似游戏的体验。

As Galloway and others have argued, the real critique of gamification rises from the logical extension of this colonial march, as gamification comes to define not just social interaction but deep structures of labor and society. One of the most compelling aspects of games is precisely the seduction of algorithmically ordered universes—spaces where our apophenia can be deeply indulged, where every event and process operates according to a rule set. These universes are aesthetically neat and tidy, with rules and conditions that, we believe, can be learned and ultimately mastered. The aesthetic of computational order echoes Bogost’s warning about the cathedral; its appeal for human engagement is rapidly expanding from play to work. Increasingly, startups are bringing this logic to the real world, creating game-like experiences for services like taxis (e.g., Uber and Lyft), household chores (Handy, HomeJoy, Mopp) and even office communications (Slack).

这些公司在设​​计企业家斯科特·贝尔斯基 (Scott Belsky) 所说的“界面层”中运营，利用有吸引力的设计将文化生活中混乱的方面澄清和合理化，使其成为简单、可靠的选择^。32

These companies operate in what design entrepreneur Scott Belsky calls the “interface layer,” using appealing design to clarify and rationalize messy aspects of cultural life into simple, dependable choices.³²

界面经济

The Interface Economy

如果说Zynga及其游戏开发者团队找到了从娱乐中榨取劳动价值的方法，那么新一波界面层公司则将劳动重新定义为一种娱乐，并借用了“共享经济”这一乐观的框架。他们的论调依赖于技术协作和即时交付的概念：利用闲置资源，例如汽车中的空座位、房屋中闲置的房间等等。但所有这些互动都植根于中介计算层，该层管理临时物流，匹配服务的买卖双方，并通过精心构建的界面构建平台访问。事实上，正如我们将在下文中看到的，最后一个因素至关重要，因此将其称为“界面经济”更为合理。在这种经济中，传统的社交和商业互动越来越多地通过依赖于复杂、严密设计的抽象和简化形式的应用程序和屏幕进行。

If Zynga and its cohort of game-makers have found ways to extract labor value from entertainment, the new wave of interface layer companies is reframing labor as a kind of entertainment, adopting the optimistic framing of the “sharing economy.” Their rhetoric relies on the notion of technological collaboration and just-in-time delivery: taking advantage of unused resources like empty seats in cars, unused rooms in houses, and so forth. But all of these interactions are grounded in the mediating computational layer that manages ad hoc logistics, matches buyers and sellers of services, and structures access to platforms through carefully constructed interfaces. Indeed, as we’ll see below, that last factor is so important that it makes much more sense to call this the “interface economy,” where traditional social and commercial interactions are increasingly conducted through apps and screens that depend on sophisticated, tightly designed forms of abstraction and simplification.

2010年代的界面经济顺理成章地延续了20世纪90年代和21世纪初第一波科技公司占据主导地位的趋势。亚马逊、Netflix和谷歌等早期巨头崛起的主要因素之一是它们能够将算法套利策略应用于既定的资本主义空间。“颠覆性”技术颠覆了我们购书、租电影和搜索信息的方式，随着这些服务转向线上，成千上万的实体店倒闭。过去几年，硅谷的孵化器和风险投资家将注意力转向了那些与传统科技领域截然不同、且随时准备进行算法重塑的全新领域。游戏化、普适计算和遥感（换句话说，万物量化）的胜利催生了一大批新兴企业，它们在此前稳定的文化空间之上添加了一层算法层面。 TaskRabbit、Uber 和 Airbnb 等公司正在采用算法逻辑来寻找住宿、交通和个人服务的新效率，在消费者和他们获取出租车、酒店和私人助理等服务的传统途径之间插入一个计算抽象层。

The interface economy of the 2010s follows logically from the first wave of technology companies to gain dominance in the 1990s and 2000s. A principal factor in the rise of early giants like Amazon, Netflix, and Google was their ability to adapt algorithmic arbitrage to established capitalistic spaces. “Disruptive” technologies upended the way we shop for books, rent movies, and search for information, shuttering thousands of brick and mortar stores as these services moved online. In the past few years, the incubators and venture capitalists of Silicon Valley have turned their attention to new areas ready for algorithmic reinvention that are more distant from the traditional technology sector. The triumph of gamification, ubiquitous computing, and remote sensing (in other words, the quantification of everything) has led to a slew of new businesses that add an algorithmic layer over previously stable cultural spaces. Companies like TaskRabbit, Uber, and Airbnb are adapting algorithmic logic to find new efficiencies in lodging, transportation, and personal services, inserting a computational layer of abstraction between consumers and their traditional pathways to services like taxis, hotels, and personal assistants.

这些公司借鉴了《FarmVille》等游戏的理念，将“近乎魔法的圆圈”强加于此前被认为是严肃业务的领域。例如，Uber 为其司机提供了一个简单的应用程序界面，让人联想到《侠盗猎车手》系列等开放空间驾驶游戏（图4.2）。该公司在定价和与司机分成比例方面的不透明性，使其更像是一款随意的电子游戏，积分是根据我们玩家仅部分理解的算法发放的。整个平台的设计旨在抽象化雇佣司机的监管和生物政治层面。员工变成了承包商，出租车队、调度员、车库和维护等既定的运营成本神奇地消失了。所有社会经济基础设施都被简单的软件界面和法律界面所掩盖，这些界面将乘客与司机连接起来，并将风险抽象为涵盖每位司机和乘客的通用综合保险单。对许多乘客来说，最吸引人的或许是支付和给小费的尴尬也被抽象化了。一旦你使用公司的应用程序叫车，支付就完全变成了后台活动，乘客在目的地下车后才会收取费用。对于大多数Uber乘车类型来说，系统内根本不允许给小费；乘客必须使用现金或某些第三方支付系统，例如无处不在的移动支付服务^Square。33

These companies take the ethos of games like FarmVille and impose their “almost-magic circle” on what was previously considered to be serious business. Uber, for example, presents a simple application interface for its drivers that is deeply reminiscent of open-space driving games like the Grand Theft Auto series (figure 4.2). The company’s opacity about pricing and the percentage of revenue shared with drivers makes it even more like an arbitrary video game where points are handed out according to an algorithm we players only partially understand. The entire platform is designed to abstract away the regulatory and biopolitical aspects of hired drivers. Employees become contractors and the established overhead of cab fleets, dispatchers, garages, and maintenance magically disappears. All the socioeconomic infrastructure gets swept away behind the simple software interfaces that connect riders with drivers, and a legal interface that abstracts risk away into generalized blanket insurance policies covering every driver and passenger. Perhaps most appealing for many riders, the awkwardness of payment and tipping is also abstracted away. Once you hail a car using the company’s app, payment becomes entirely a background activity, with charges applied once the rider exits the vehicle at her destination. For most Uber ride types tipping is not possible within the system at all; passengers must use cash or some third-party payment system, such as the ubiquitous mobile payment service Square.³³

Uber 在支付流程上强加的抽象层很能说明问题。事实上，Uber 的客户，也就是游戏中的玩家，既是司机，也是乘客，因为公司通过撮合这两类群体来收取佣金。我们在广告中也看到了这种安排，比如图 4.3中，时尚的黑白摄影将走出租来的车的奢华体验浪漫化，同时将穿着考究的司机和乘客置于财富和成就的同一层面。“掌控当下”这句话用在一家像所有游戏化供应商一样依赖于时间抽象和规则的公司身上再合适不过了。如果说《FarmVille》迫使玩家在白天和晚上的任意时间照料庄稼，那么 Uber 则承诺将玩家从其他系统的时间霸权中解放出来。作为乘客，不再需要为找出租车或等出租车而焦虑或不确定；作为司机，不再需要与调度主管纠结于何时何地可以开车。这里的界面层在即时时间和距离方面提供了某种形式的确定性——可用乘客和司机的卡通地图视图的近乎魔法圈巧妙地描绘了这一点。但它也带走了很多东西：对 Uber 决定车费的财务模型的全面理解、客户奖励卓越服务的机构，以及对出租车和出租车服务的既定监管计划的参与（这让 Uber 陷入了世界各地的法律纠纷）。

The layer of abstraction Uber imposes over the payment process is telling. In fact, Uber’s customers, the players in its game, are both the drivers and the riders, since the company collects its commission based on bringing these two groups together. We see the arrangement in advertisements like the one in figure 4.3, where the sleek black and white photography romanticizes the luxury of emerging from a hired car even as it puts the driver and the rider, both wearing elegant clothes, on the same plane of wealth and accomplishment. “Owning the moment” is a fitting phrase for a company that, like all vendors of gamification, depends on the abstraction and regulation of time. If FarmVille forces its players to tend their crops at arbitrary times of the day and night, Uber promises liberation from the temporal hegemony of other systems. As a rider, there is no more anxiety or uncertainty about locating or waiting for a cab; as a driver, no more struggling with a dispatch supervisor about where or when one is allowed to drive. The interface layer here provides certain forms of certainty in terms of immediate time and distance—neatly pictured by the very literal almost-magic circle of the cartoon map views of available riders and drivers. But it also takes many things away: a complete understanding of the financial model by which Uber decides how much a fare will cost, the customer’s agency to reward exceptional service, and engagement with established regulatory schemes for taxis and livery services (which has landed Uber in legal battles around the world).

Uber 和其他界面层公司的浪漫情怀源于将个人理想化为一人公司——一个由独立 CEO 组成的国家，他们随时随地、随心所欲地工作。Uber 只是将这一界面层融入众多不同文化空间的更广泛运动中的一个突出例子，这些空间涵盖从雇佣房屋维修承包商到促进私人派对汽车销售等各种领域。当然，所有这些市场都已技术化，但在智能手机和无处不在的传感器出现，能够密切监控人力和财务资源之前，它们基本上无法直接进行算法管理。就劳动力和剩余价值而言，Uber、Airbnb 及其同类算法所利用的是现代消费中闲置的基础设施：空置的汽车、闲置的卧室和就业不足的人群。根据加州大学洛杉矶分校城市规划研究员唐纳德·舒普 (Donald Shoup) 的研究，平均而言，汽车 95% 的时间都处于停放状态；为什么不利用这种潜在资源呢？³⁴

The romance of Uber and other interface layer companies depends on the idealization of the individual as a one-person corporation—a nation of independent CEOs working where, when, and how they please. Uber is merely one prominent example of the broader movement to build this interface layer into many different cultural spaces, from hiring contractors for home repair to facilitating private party car sales. All of these markets were, of course, already technological, but they were largely inaccessible to direct algorithmic management until the advent of smartphones and ubiquitous sensors enabling the close monitoring of human and financial resources. In terms of labor and surplus value, what the algorithms of Uber, Airbnb, and their cohort capitalize on is the slack infrastructure of modern consumption: empty cars, unused bedrooms, and under-employed people. According to UCLA urban planning researcher Donald Shoup, the average car is parked 95 percent of the time; why not exploit that latent resource?³⁴

从更广泛的角度来看，界面层是对当代资本主义静水区的殖民——对已被消费或部署的商品和空间进行重新调动。最终，这些系统与游戏化一样，对商品、人类注意力和时间进行套利，激励我们创造新的经济效率并从中获取收益。另一家名为 Yerdle（致力于通过一种算法交换市场回收不需要的消费品）的初创公司联合创始人对此做出了恰如其分的阐述：“我们希望让人们把事情做得更好。” ³⁵这种通过资本主义激励积极效率的利他主义抱负与谷歌首席执行官埃里克·施密特的观点产生了共鸣，施密特认为，客户“希望谷歌告诉他们下一步应该做什么”。这是算法游戏化与市场之间实施断层线的核心劳动力问题：谁在推动这些变化，以及我们在共享经济中究竟“共享”了什么？

Viewed more broadly, the interface layer is a colonization of the quiet backwaters of contemporary capitalism—the remobilization of goods and spaces after they have already been consumed or deployed. Ultimately these systems engage in precisely the same kinds of arbitrage of goods, human attention, and time that gamification does, motivating us to create new economic efficiencies and extract revenue from them. The cofounder of another startup, named Yerdle (dedicated to recycling unwanted consumer goods through a kind of algorithmic swap-bazaar), put it just right: “We want to make people make things better.”³⁵ This altruistic ambition to motivate positive efficiencies through capitalism resonates with Google chief Eric Schmidt’s suggestion that customers “want Google to tell them what they should be doing next.” This is the central labor question at the implementation fault line between algorithmic gamification and the marketplace: who is motivating these changes, and what exactly are we “sharing” in the sharing economy?

从最明显的层面来看，这种新经济关乎更高效地获取私有或原子化的商品和服务。像Yerdle和Airbnb这样的公司的宣传语依赖于物质资源的调动：汽车、公寓以及闲置的家居用品。分享你的个人物品，将闲置的物品货币化，减少拥有这些物品的开销，将空置的车辆或房间变成利润中心和社区资源。从更深层次来看，界面创业者要求我们分享（并货币化）我们的时间：Lyft的创始人不仅受利润驱动，还被困在车里的普通通勤者的孤独感所驱动。36^这些公司鼓励我们将时间奉献给他人，其诉求通常将劳动报酬的诱惑与更复杂的社会关系融合在一起。Uber向其司机和乘客兜售一种精英独立性（图4.3），而Lyft则兜售一种不同且更亲密的社交联系（图 4.4）。该公司最近才取消了要求司机用古怪的粉红色胡须装饰汽车的规定，许多司机仍然认为乘客会舒适地坐在前排座位上，而不是后排座位上。

On the most obvious level, this new economy is about more efficient access to privately owned or atomized goods and services. The rhetoric of companies like Yerdle and Airbnb leans on the mobilization of material resources: cars, apartments, and household objects that are sitting around unused. Share your personal goods to monetize that slack and reduce the overhead of ownership, turning an empty vehicle or room into a profit center and a community resource. At a deeper level, what the interface entrepreneurs are asking is for us to share (and monetize) our time: the founders of Lyft are motivated not just by profit but by the loneliness of the average commuter stuck in his car.³⁶ These companies encourage us to dedicate our hours to others, often in appeals that blend the allure of wages for labor with something more socially complex. Where Uber sells a kind of elite independence to both its drivers and riders (figure 4.3), Lyft is selling a different and more intimate kind of social contact (figure 4.4). The company only recently abandoned its directive that drivers festoon their cars with quirky pink moustaches, and many drivers still assume passengers will sit companionably in the front seat, rather than the rear.

对于像Lyft这样的公司，以及像约会应用Tindr这样刻意打造亲密界面层的系统来说，“共享经济”根本不关乎金钱，而关乎陪伴的体验。如果这些商业模式建立在剥削某些异化的劳动和注意力的基础上，那么它们的客户体验有望缓解这种异化。即使Uber和Lyft从不可见的交易中收取隐形佣金，其情感体验也是一种特殊品牌社区的体验。但这个社区至关重要，本质上是由算法来调解的。司机和乘客通过计算进行评级和审查；将他们联系在一起的界面层也是信任的核心仲裁者。难怪对这些公司最严重的威胁不是财务丑闻，而是对这种信任的攻击，例如Uber被曝追踪记者的行踪，或其司机因性侵乘客而被捕的事件。³⁷共享经济最终依赖于一种原子化的亲密关系，即与陌生人的一系列短暂而亲密的接触，由算法文化机器进行管理和承保（在情感、财务和责任方面）。

For companies like Lyft and more deliberately intimate interface layer systems like the dating app Tindr, the “sharing economy” is not about money at all, but about that experience of companionship. If these business models are founded on exploiting certain kinds of alienated labor and attention, their customer experience promises relief from that alienation. Even as Uber and Lyft collect their invisible commissions on unseen transactions, the affective experience is one of a specially branded community. But that community is crucially, essentially mediated by the algorithm. Drivers and riders are rated and vetted through computation; the interface layer bringing them together is also the central arbiter of trust. Little wonder that the most serious threats to these companies are not financial scandals but attacks on that trust, as when Uber was revealed to be tracking the movements of journalists, or incidents where its drivers have been arrested for sexually assaulting passengers.³⁷ The sharing economy ultimately depends on an atomized form of intimacy, a series of fleeting, close encounters with strangers that are managed and underwritten (in emotional, financial, and liability terms) by algorithmic culture machines.

虽然这种亲密感对于共享经济的运作至关重要，但它并非这些系统销售的主要商品。真正的卖点在于界面本身：以计算为媒介的文化体验及其底层算法的简化和抽象。一个素未谋面的陌生人来打扫你的浴室，或者坐在你旁边开车送你回家，这种感觉既亲密又尴尬。更具吸引力的是按需提供服务的界面的美感。算法以经过校准的情感运行，并提供一种封装的体验，这种体验通常被用来让那些对鲜明的阶级划分感到不适的消费者更容易接受这些尴尬的交易。许多界面公司实际上将自己定位为现有服务的“包装器”，将它们捆绑、组织并去神秘化，以提供轻松的用户体验。但这种包装器既是体验性的，也是物流和商业性的，它用共享经济的价值观重新构建了原始服务或产品的语境。我们感受到的亲密感受到底层界面的精心约束。通过智能手机的匿名性和远程性，给女佣打一星比解雇自己雇佣的人要容易得多。这些系统还会通过奖励技术素养来强化阶级分化，使数字经济的解放更多地惠及那些拥有教育背景、金钱和时间掌握这些技能的用户。

While this intimacy is necessary for the sharing economy to function, it is not the primary commodity these systems are selling. The real sell is the interface itself: the experience of computationally mediated culture and its underlying algorithmic simplification and abstraction. Having a stranger you have never met arrive to clean your bathroom or sit next to you while they drive you home is intimate but also awkward. Much more appealing is the aesthetic of the interface that delivered that service on demand. The algorithm performs with a calibrated affect, and delivers an encapsulated experience that is typically put to the use of making these awkward transactions more palatable for consumers who find stark class divisions uncomfortable. Many interface companies literally present themselves as “wrappers” around existing services, bundling, organizing, and demystifying them for a painless user experience. But the wrapper is experiential as well as logistical and commercial, recontextualizing the original service or product with the values of the sharing economy. The intimacy we feel is carefully constrained by the underlying interface. Far easier to give that maid one star through the anonymity and distance of the smartphone than to fire someone you hired yourself. These systems can also reinforce class divides by rewarding technoliteracy, making the emancipation of the digital economy disproportionately available to users with the education, money, and time to master those games.

当阶级的幽灵真的出现时，界面经济暗示我们都是共同的用户——同一个临时经济区域的所有居民。像Lyft这样的公司让我们相信，我们可以是司机，也可以是乘客；可以是承包商，也可以是服务消费者；而这些体验是由本质上“与我们相似”的人提供的。界面层充当着挡板或减震器的作用，通过五星评级和计算验证的信任的魔力，消解了阶级、性别和种族等常见的社会经济障碍。这种套利形式是对亲密关系的复杂三角测量，要求参与者交换或重新协商某些身份特征，以换取算法精心设计的替代品。长期以来，在纽约街头叫出租车的人一直感到自己受到种族等特征的评判，而一些非裔美国人之所以选择优步，正是因为种族背景在叫车过程中被淡化或消除了。³⁸ Uber 成为社会背景、身份政治、监管制度和反歧视法体系中的一个新的过滤器，通过其界面的视角来影响这些形形色色的意识形态冲突。正如华盛顿特区一位博主所言：“对非裔美国人来说，Uber 的体验实在是太便捷了。没有争吵，也没有对话。我需要车的时候，它就来了。它会把我送到目的地。虽然我必须为这种体验支付额外费用，但这一切都是值得的。” ³⁹

When the specter of class does appear, the interface economy suggests that we are all users together—all denizens of the same ad hoc economic zone. Companies like Lyft convince us that we can be either the driver or the rider, the contractor or the consumer of services, and that these experiences are being provided by people fundamentally “like us.” The interface layer acts as a baffle or shock absorber, diffusing common socioeconomic barriers like class, gender, and race through the magic of five-star ratings and computationally validated trust. This form of arbitrage is a sophisticated triangulation of intimacies, asking participants to trade in or renegotiate certain facets of identity in favor of algorithmically crafted substitutes. People hailing cabs on the streets of New York have long felt judged by race, among other characteristics, and some African Americans specifically use Uber because racial context is muted or removed from the ride-hailing equation.³⁸ Uber becomes a new filter in the system of social context, identity politics, regulatory regimes, and antidiscrimination laws, inflecting these various ideological conflicts through the lens of its interface. As one blogger in Washington, DC, noted, “The Uber experience is just so much easier for African-Americans. There’s no fighting or conversation. When I need a car, it comes. It takes me to my destination. It’s amazing that I have to pay a premium for that experience, but it’s worth it.”³⁹

简而言之，界面经济的基本商品是访问这些系统的权限——与其说是它们提供的服务，不如说是计算结构本身，而这些结构反过来又映射或模拟着日益丰富的文化和社会空间。在劳工政治的语境下，界面经济将商业交易原子化，使其脱离空间和社会语境，并将城市的文化地理转化为一个或另一个抽象的空间，如同小屏幕上的卡通地图，却又掩盖了它所代表的真实城市。我们越来越多地将偏见、商业可行性，乃至语境本身的判定外包给算法，要求界面层不仅为我们完成后勤工作，还为我们完成伦理和文化工作。所有这些“共享”确实提供了新的亲密关系和联系形式，消除或减少了某些传统的偏见形式，但它需要大量的前期投入。为了成为算法界面中的经济参与者，我们必须接受计算信任系统，该系统通过我们几乎无法察觉的抽象和套利方程式来计算价值。在简便、友好的计算界面背后，是一个劳动力转化、补救和剥削的世界。

In short, the fundamental commodity of the interface economy is access to these systems—not the services they provide so much as the computational constructs themselves, which in turn map or simulate a growing range of cultural and social spaces. In the context of labor politics, the interface economy atomizes commercial transactions, cutting them out of spatial and social context, and converts the cultural geography of the city into one or another abstracted space, a cartoon map on a small screen that nevertheless obscures the real city it represents. Increasingly we outsource the determination of bias, of commercial viability, indeed of context itself, to the algorithm, asking the interface layer to do not just the logistical but the ethical and cultural work for us. All of this “sharing” does offer new forms of intimacy and connection, eliminating or reducing certain traditional forms of bias, but it demands a steep investment up front. In order to become economic actors in the algorithmic interface, we have to buy in to the computational trust systems that calculate value through equations of abstraction and arbitrage that are largely invisible to us. Behind the facade of the facile, friendly computational interface, there is a world of labor translation, remediation, and exploitation at work.

算法劳动：云仓库

Algorithmic Labor: Cloud Warehouses

虽然界面经济支撑着许多人与人之间的互动，消费者和承包商在相对平等的基础上运作，但算法套利的另一个领域几乎完全被计算的面纱所掩盖。颠覆传统蓝领行业的颠覆性创新改变了许多公司对其业务运营的基本假设。我们目前讨论的所有算法企业——例如Zynga、Uber、谷歌和苹果——都依赖于庞大的分布式全球计算基础设施：数百万台服务器在巨大的仓库中全天候运行，处理和管理着海量的数据存储。这种数据“云”是文化机器与算法过程之间抽象层的另一个关键环节，它使得我们所依赖的所有神奇功能成为可能，例如Siri的快速响应或谷歌搜索结果的即时性。克里斯蒂安·桑德维格 (Christian Sandvig) 指出了云这一隐喻的迷人历史，它最初是 20 世纪 70 年代计算机工程师在系统流程图中使用的一个图标：“这一术语来自网络图的符号体系，其中云符号表示图中与其内部细节无关的部分。” ⁴⁰因此，根据定义，云是一种抽象，是一种将系统中不太有趣的方面括起来的方式。

While the interface economy supports many human-to-human interactions where consumers and contractors operate on relatively equal footing, there is another sphere of algorithmic arbitrage where human labor is almost entirely obscured behind the veil of computation. The disruptive innovations that upended traditional blue-collar industries have changed the fundamental assumptions of many corporations around their business operations. All of the algorithmic businesses we have discussed so far—companies like Zynga, Uber, Google, and Apple—depend on a massive infrastructure for distributed, global computation: millions of servers running around the clock in vast warehouses, processing and managing tremendous data stores. This “cloud” of data is another crucial link in the layers of abstraction between culture machines and algorithmic process, enabling all of the magic tricks we have come to depend on, like the speedy responses of Siri or the immediacy of search results from Google. Christian Sandvig has pointed out the fascinating history of the cloud metaphor, which began life as an icon computer engineers in the 1970s would use in system flowcharts: “The term comes from the symbology of the network diagram, where a cloud symbol indicates a part of the diagram whose internal details are irrelevant.”⁴⁰ By definition, then, the cloud was an abstraction, a way to bracket off less interesting aspects of a system.

如今，我们用云的隐喻来描述原子和比特的管理，承诺随时随地即时访问信息和商品。现实世界中杂乱无章的物流要求像亚马逊和苹果这样的公司管理或处理庞大的仓库、工厂和运输网络。这些设施配备了数百万工人，他们在为计算云和比特与原子的疯狂交换提供服务时，受制于工作场所的算法逻辑。正如环境学者艾莉森·卡鲁斯 (Allison Carruth) 所说，幕后的庞大服务器群是“能源密集型、大规模工业化的基础设施”，在云的隐喻朦胧、虚幻的美感背后隐藏着巨大的生态成本。41^这些地方是界面层的工厂、仓库和套利点，人力、金融、信息和电能的储备在这里被转换并路由到各自的目标。

Today we use the cloud metaphor to describe the management of atoms as well as bits, promising instant access to information and goods anywhere, at any time. The messy logistics of physical reality require companies like Amazon and Apple to manage or transact with vast warehouses, factories, and transportation networks. These facilities are staffed by millions of workers who are beholden to an algorithmic logic of the workplace as they service the computational cloud and the frenetic interchange of bits and atoms. As environmental scholar Allison Carruth put it, the vast server farms behind the scenes are “an energy-intensive and massively industrial infrastructure” that hides major ecological costs behind the hazy, insubstantial aesthetic of the cloud metaphor.⁴¹ Such places are the factories of the interface layer, the depots and arbitrage points where reserves of human, financial, informational, and electrical energy are translated and routed to their various targets.

云设施种类繁多，从纯粹的数据档案库到字面意义上堆满物品的仓库。其中最具体、最接近我们熟悉的劳动力和资本形式的，是云仓库，它们满足我们对数百万消费品隔夜或当日送达的需求。像亚马逊这样的公司会与第三方物流公司签约，为这些设施提供服务，并雇佣他们的员工，他们的职位名称是“拣货员”。拣货员就像人脑计算机，做着机器人无法完成的复杂工作：在平板电脑的指挥下，他们在数千个货箱中定位物品，并将库存物料转化为算法流程。每一步、每一秒都被追踪，他们的表现和持续工作取决于是否达到特定的、严格限制的目标。42^仓库是算法逻辑的物理表达：一个特定的货箱可能存放着随机排列的物品，以同样随意的方式摆放在过道里。拣货员可能会将一件物品放在上面几秒钟，然后将其放在传送带上，由设施中其他地方的另一个人进行打包。这些临时员工职业生涯的方方面面都由算法套利决定：雇佣的工人数量和分配的轮班（通常在工作开始前几小时或几分钟才确定）；每个工人的生产力目标，以秒、步数和出货量计算；迟到、谈话甚至工作场所温度的政策（在夏季热浪期间，人们认为在亚马逊的一个仓库排队等候救护车比安装空调更好或更有效率）^。43

There are many kinds of cloud facilities, from pure data archives to very literal warehouses full of stuff. The most tangible of these, the closest analog to familiar forms of labor and capital, are the cloud warehouses that supply our needs for overnight or same-day shipping on millions of consumer products. Companies like Amazon contract with third-party logistics outfits to service these facilities and hire their workers, who have job titles like “picker.” Pickers are human computers doing work just a little too complex for robots: directed by tablet computers to locate items among thousands of bins, they translate standing reserves of material into algorithmic processes. Every step and second is tracked, and their performance and continued employment depend on meeting specific, tightly constrained targets.⁴² The warehouses are a physical expression of algorithmic logic: a particular bin might hold a random assortment of objects, sitting in an aisle with similarly haphazard organization. The picker might hold an object for a few seconds before putting it on a conveyer belt to be packaged by another human elsewhere in the facility. Every aspect of these temporary employees’ professional lives is dictated by algorithmic arbitrage: the numbers of workers hired and the shifts assigned (often determined only hours or minutes before work is to start); the productivity goals of each worker, calculated in seconds, steps, and units shipped; the policies for lateness, conversation, and even workplace temperature (during a summer heat wave, it was deemed better, or more efficient, to line up ambulances at one Amazon warehouse than install air-conditioning).⁴³

这些条件极其不人道，也只有机器才能理解其逻辑。雇佣卧底记者麦克·麦克莱兰的第三方物流公司对员工入职第一周迟到实行零容忍政策。一位员工的妻子在他入职第一周就生了孩子，结果他因为缺勤一天而被立即解雇，不得不重新走一遍招聘流程，几周后才能重返公司。除了这些政策带来的情感压力和经济压力之外，从人类运营管理的角度来看，它们也毫无意义（在前面的例子中，这包括重复昂贵的招聘文书工作、药物测试和背景调查）。但算法文化的上层建筑通过使其变得无形，使这种无意义变得合理。云看似无形，却掩盖或模糊了计算奇观背后的这些抽象层和人力劳动，向我们传递着关于无摩擦电子商务的诱人理想、无与伦比的价格和快速的免费送货的信息。

The conditions can be deeply inhumane and logical only to machines. The third party logistics company that hired undercover reporter Mac McClelland had a zero tolerance policy for lateness in the employee’s first week. One employee whose wife delivered a baby in his first week was summarily fired for missing a day’s work, requiring him to repeat the hiring process and rejoin the company a few weeks later. Beyond the emotional duress and financial hardship of such policies, they make little sense from a human operational management perspective (in the preceding example, this would include duplicating costly hiring paperwork, drug tests, and background checks). But the superstructure of algorithmic culture makes sense of this senselessness by rendering it invisible. The cloud seems physically intangible even as it serves to cover or obscure these layers of abstraction and human labor behind the spectacle of computation, showering us with messages about the seductive ideal of frictionless e-commerce with unbeatable prices and fast free shipping.

对于麦克莱兰和她的同事们来说，算法抽象的空间已经完全吞噬了现实——拣货员的工作本身就充满了随意操控的规则，但失败的后果却要严重得多。仓库中“近乎魔幻的圆圈”颠覆了我们在娱乐和媒体游戏化中看到的现实与虚拟的关系。不同于Uber的魔力，一辆卡通地图上的汽车在现实生活中出现在路边，云端工作者仍然被困在算法的卡通世界中，只有在他们不可预测的轮班结束时才能逃脱。这些工作者维护着支撑资本主义科幻幻想的基础设施：一键购买、无限选择和近乎即时满足的无摩擦商业。但对于服务于这些仓库的工人来说，步数统计、数字目标和其他游戏机制的架构体现了这种理想化、游戏式叙事在现实空间和现实生活中的现实。云就像一层不透明的膜，转移了公众的注意力、道德调查和法律责任，因此我们都可以继续玩网上购物游戏，而没有义务，也很少有机会去同情或与人互动。

For McClelland and her coworkers, the space of algorithmic abstraction has entirely subsumed reality—the game that is the picker’s job still comes with arbitrary, manipulative rules, but the consequences of failure are much higher. The “almost-magic circle” of the warehouse reverses the relationship between real and virtual that we see in the gamification of entertainment and media. Unlike the magic of, say, Uber, where a car from a cartoon map materializes in real life at the curb, cloud workers remain trapped in the cartoonish world of the algorithm, escaping only at the end of their unpredictable shifts. These workers maintain the infrastructure supporting a capitalist, science fiction fantasy: the frictionless commerce of one-click purchasing, limitless choice, and near-instant gratification. But for the workers servicing these warehouses, the architecture of step-counting, numerical targets, and other game mechanics embody the realities of that idealized, game-like narrative across real space and real lives. The cloud functions as an opaque membrane deflecting public attention, ethical inquiries, and legal liability, so we can all continue playing the online shopping game with no obligation, and few opportunities, to empathize or interact with a human being.

对于这个圈子里的员工来说，没有什么魔法可言，除非你算上奥威尔式的传感器力量。这些传感器无情地测量你的行为，强加一个外部化的数字价值指数，比如浮在每个员工头顶的高分或健康条，但通常只有管理层才能看到。就他们与全球算法系统的权力关系而言，美国云计算员工的生活与众所周知的苹果承包商富士康在中国的大型工厂的苦难并无太大区别。44^那里的工人，有些年仅14岁，一直忍受着极长的工作时间、危险的工作条件和其他严重问题的折磨。45该公司董事长郭台铭在²⁰¹²年直言不讳：“[我们]在全球拥有超过一百万的员工，而人类也是动物，管理一百万只动物让我头疼。” ⁴⁶到目前为止，企业仍然无法摆脱威廉·吉布森曾经称之为“肉体空间”的东西，即生物物质的柔软领域。生物动力推动着界面经济的许多关键方面，尽管机器人技术和人工智能的进步有朝一日可能会取代这些“动物”所从事的大部分工作。除了斥资 7.75 亿美元收购工厂自动化公司 Kiva Systems 之外，亚马逊还在测试机器人手臂和其他自动化系统，这些系统可能会彻底取代人类的拣货员工作^。47

For the workers inside this circle there’s no magic to speak of unless you count the Orwellian power of sensors that relentlessly measure your behavior, imposing an externalized, numerical index of worth like a high score or a health bar floating over each worker’s head, but often visible only to management. In terms of their power relationship to global algorithmic systems, the lives of American cloud workers are not that different from the well-known miseries of Apple contractor FoxConn’s massive facilities in China.⁴⁴ There workers, some children as young as fourteen, have persistently suffered under incredibly long hours, dangerous working conditions, and other serious problems.⁴⁵ Terry Gou, the company’s chairman, put it bluntly in 2012: “[We have] a workforce of over one million worldwide and as human beings are also animals, to manage one million animals gives me a headache.”⁴⁶ So far, corporations have been unable to escape what William Gibson once called meatspace, the squishy realm of biological materiality. Biopower fuels many crucial aspects of the interface economy, though advances in robotics and artificial intelligence may one day eliminate the jobs most of these “animals” hold. In addition to spending $775 million to acquire the factory automation firm Kiva Systems, Amazon is already testing out robot arms and other automated systems that could eliminate the human job of picker entirely.⁴⁷

正如亚马逊当日送达服务对物流的严格要求一样，富士康的条件也由苹果与其消费者的关系支撑。我们预计该公司将在极短的时间内（有时仅需数月）快速开发、制作原型，并量产数百万台设备，同时保持近乎完全的保密性，直到新产品能够引人注目地亮相。我们期待全球发布日期、不断提升的性能和创新，以及来自库比蒂诺设计奇才的即时满足感。所有这些都是可能的，而且利润丰厚，但只有通过由算法操控的工厂车间、工人宿舍和掠夺性做法等物流和人力基础设施才能实现。

Like the intense logistical requirements of Amazon’s same-day delivery services, FoxConn’s conditions are underwritten by Apple’s relationship with its consumers. We expect the company to rapidly develop, prototype, and then mass-produce millions of devices over incredibly short spans of time—sometimes mere months—while maintaining near-total secrecy until the new product can be dramatically unveiled. We expect global release dates, ever-increasing performance and innovation, and instant gratification from the design wizards of Cupertino. These things are all possible, and highly profitable, but only through the logistical and human infrastructure of algorithmically mediated factory floors, worker dormitories, and predatory practices.

云仓库和云工厂将计算逻辑转移到物理空间；跨越抽象与实现之间的鸿沟意味着使它们成为人类工作环境的糟糕场所。算法套利从数百万报酬微薄的人工劳动原材料中提取即时性、相关性和快速创新。这些系统依靠复杂的后端，通过云的魔力来推广算法文化的浪漫，根据计算逻辑无形地组织数百万人类并执行命令，以营造无缝、轻松高效的印象。这些系统确实高效，但主要并非体现在我们作为消费者身上。亚马逊、苹果和其他公司不断寻找算法解决方案，以便更快地向我们提供商品和服务，但它们真正的工作，即成功的文化套利，是通过盈利来实现的。

Cloud warehouses and factories transpose computational logic to physical spaces; spanning the gap between the abstract and the implemented entails making them terrible places for humans to work. Algorithmic arbitrage extracts immediacy, relevance, and rapid innovation from the raw materials of untold millions of poorly compensated person-hours of labor. These systems rely on a sophisticated backend to promote the romance of algorithmic culture through the magic of the cloud, invisibly organizing millions of humans according to computational logic and executing orders to support the impression of seamless, effortless efficiency. And the systems are efficient, but not primarily in the sense presented to us as consumers. Amazon, Apple, and other companies continue to find algorithmic solutions to deliver goods and services to us ever faster, but their real work, the winning cultural arbitrage, is in doing so at a profit.

我们几乎可以想象，把这条神奇的管道反过来会是什么样子。与其说它是一个按需交付宏伟技术奇迹的系统，不如说它为我们带走了什么。它吸收了现实世界的所有尴尬和麻烦，像倾倒有毒废物一样倾倒在云端工作者身上，而他们必须围绕抽象的、算法化的就业和价值结构来安排自己的生活。它吸走了所有浪费在排队、等待、在商场停车场里缓慢挪动车辆的时间……并将这些时间、时间甚至时间埋葬在寒冷的服务器群的冷藏室里，在云端呼呼作响的风扇和等待的CPU周期中无休止地循环。

One can almost imagine putting the magical pipeline in reverse. Instead of a system that delivers magnificent technological wonders to us on demand, think about what it whisks away for us. It suctions up all the awkwardness and hassle of the real world, dumping it like so much toxic waste onto cloud workers who must organize their lives around abstract, algorithmic structures of employment and value. It vacuums all the wasted time of standing in line, waiting on hold, inching cars around parking lots at the mall … and entombs those seconds, minutes, and years in the cold storage of frigidly air-conditioned server farms, endlessly recirculated by the whirring fans and waiting CPU cycles of the cloud.

人工智能

Artificial Artificial Intelligence

算法劳动的最终目标是将物理和文化基础设施完全抽象化。云端的工厂、服务器群和仓库仍然需要庞大的实体厂房，但一些项目致力于将算法经济逻辑更进一步。如果界面创业者试图将高度专业化的业务从其原有的语境中解放出来，例如出租车和家政服务，那么亚马逊的Mechanical Turk则走得更远，试图在云端创建一个新的通用工业基础——由一群执行各种任务的工人组成的集合，所有工人都独立工作，由算法管理。亚马逊颇具讽刺意味地将其称为“人工人工智能”，这巧妙地概括了其对人机关系的重塑。该系统创建了一种与人类思维的接口——一个工业化的“nam-shub”（人脑接口），可以快速利用成千上万人的脑力来运行特定的程序。这里的单元操作是高度具体的“人类智能任务”（HIT），通常只需几秒钟即可完成：例如“拨打一个号码，记下他们的回答”、“从购物车收据中提取购买的物品”或“旋转 3D 模型以匹配图像”。48 ^Mechanical Turk 不会将简单的重复性任务分配给计算机器，而是将它们分配给通常在每个 HIT 上赚取几美分的个人。49^从字面意义上讲，他们只是作为计算文化机器的技术扩展而运作的人类。在系统界面中，这些人是匿名的，仅通过字母数字代码（图 4.5），他们的身份由他们的绩效记录和一套“资格”来定义，这些资格将一些工人提升为照片审核和分类等任务中的“大师级工人”的称号^。50

The final telos of algorithmic labor is the work that abstracts physical and cultural infrastructure away altogether. The factories, server farms, and warehouses of the cloud still require massive physical plants, but some projects strive to take the algorithmic economy logic a step further. If the interface entrepreneurs seek to unmoor highly specific businesses from their original contexts, like taxis and house cleaning services, Amazon’s Mechanical Turk goes much farther in its attempt to create a new general industrial base in the cloud—an assemblage of workers for a huge range of tasks, all working on their own, managed by algorithm. Amazon rather ironically bills this as “artificial artificial intelligence,” which neatly summarizes its reinvention of the human–machine relationship. The system creates a kind of interface to the human mind—an industrial nam-shub—for quickly harnessing the brain power of thousands of people to run a specific program. The unit operations here are highly specific “human intelligence tasks” (HITs) that often take only seconds to complete: “call a number, note how they answer,” “extract purchased items from a shopping cart receipt,” or “rotate 3D model to match image,” for example.⁴⁸ Instead of assigning simple repetitive tasks to a calculating machine, Mechanical Turk assigns them to individual humans who typically make a few cents on each HIT.⁴⁹ They are, in a very literal sense, humans functioning as mere technical extensions of a computational culture machine. Within the system interface these people are anonymized and identified only by an alphanumeric code (figure 4.5), their identities defined by their performance records and a set of “qualifications” that elevate some workers to the title of “master worker” in tasks like photo moderation and categorization.⁵⁰

该系统的名称源自人工智能神话中的另一个开创性时刻，即沃尔夫冈·冯·肯佩伦 (Wolfgang von Kempelen) 于 1770 年揭幕的“土耳其人”自动机，令中欧宫廷惊叹不已。这是一个骗局，由一位精心隐藏在齿轮和显示面板柜子里的、非常人性化的国际象棋大师控制，但它是一个强大的奇观：一个可以击败几乎任何对手的机械思维机器。土耳其人被描述为一个为游戏而设计的工作机器，一个计算引擎，它的承诺直到两个世纪后才兑现，即 1997 年 IBM 的深蓝击败了国际象棋世界冠军加里·卡斯帕罗夫。原始土耳其人的历史阐明了亚马逊为这种分布式人机计算系统命名的文化力量，以及人工智能的修辞力量。在1770年和1997年，算法计算都以非常人性化的方式被运用来解决人类的问题，完成着一种除了以人类的眼光来看，如同游戏或智力竞赛般毫无意义的工作。在这两种情况下，服务员都推出一些设备，这些设备实际上是输入和输出受到严格限制的黑匣子，它们在备受瞩目的挑战赛中的表现，既关乎计算的文化形象，也关乎棋步的计算。这些“思考机器”执行着人类最严格、最广泛应用的分析思维形式（或游戏）。机器思考——这本身就是一场奇观，起初是欺诈性的，后来遭到激烈的质疑（至少卡斯帕罗夫是这么认为的），而后在第二次挑战中被广泛接受。

The system draws its name from another seminal moment in the mythos of artificial intelligence, the “Turk” automaton that dazzled the courts of central Europe at its unveiling in 1770 by Wolfgang von Kempelen. It was a hoax controlled by a carefully concealed, all-too-human chess master within its cabinet of gears and display panels, but it was a powerful spectacle: a mechanical thinking machine that could defeat almost any opponent. The Turk was presented as a working machine designed for play, a calculating engine whose promise was only fulfilled two centuries later when IBM’s Deep Blue defeated chess world champion Gary Kasparov in 1997. The history of the original Turk illuminates the cultural force of Amazon’s name for this distributed human computation system and the rhetorical power of artificial artificial intelligence. In both 1770 and 1997, algorithmic calculations were harnessed in a very human way for a human problem, doing a form of work that is meaningless except in the human terms of a game or contest of wits. In both cases attendants wheeled out devices that were literal black boxes with highly constrained inputs and outputs, and their performance in highly publicized challenge matches was as much about the cultural figure of computation as it was about computing chess moves. These were “thinking machines” performing one of humanity’s most rigorous and widely established forms (or games) of analytical thought. Machines thinking—that was the spectacle, fraudulent in the first instance, vehemently contested (by Kasparov, at least) and later widely accepted in the second.

土耳其人对机械思维的热情投入并非新鲜事，但其智能却令人耳目一新：据历史学家杰西卡·里斯金 (Jessica Riskin) 称，自动机至少两百年前就已在欧洲的教堂和宗教节日中盛行。51^在宗教改革及其对神性和人为神器的严格区分之前，信众们欣赏着“机械而神圣”的机器奇观^。52它们的表演充满智慧，既能激发幽默感，又能激发虔诚。然而，魔力依然存在于博克斯利修道院的恩典十字架等作品中。16 世纪，修道院每年都会用一个会动会转白眼的基督像来迎接朝圣者。人类工匠和表演者对这些表演至关重要，他们模拟反应能力和意识，从而有效地吸引观众。这种智慧的奇观在卡斯帕罗夫与深蓝的比赛中有所体现，包括他坚持不懈地部署能够突破 IBM 算法的开局和下法。⁵³人类工程师从未完全隐藏在机制背后，IBM 员工在每场比赛之间都会对系统进行调整。这种关系的最好例子或许是第一局比赛结束前一步备受关注的着法，其真相直到 2012 年 IBM 研究员 Murray Campbell 接受著名统计学家 Nate Silver 采访时才得以揭晓。这步着法让卡斯帕罗夫“心烦意乱”，因为它似乎反映了人类智力的模糊性和精细性，许多人认为这步着法分散了大师对第二局比赛的注意力，最终他输掉了比赛。54^事实上，正如 Campbell 所揭示的，这步着法是一个 bug，工程师在第一场比赛后就纠正了它。55^{卡斯帕罗夫}本人利用算法创造了奇迹，为最终只是随机计算产物的东西发明了一个复杂的文化解释。

The Turk’s spring-wound embrace of mechanical thought was not a complete novelty, though its intelligence was: automata had pervaded the cathedrals and religious festivals of Europe for at least two hundred years, according to historian Jessica Riskin.⁵¹ Before the Reformation and its stern separation of divinity and human artifact, congregations enjoyed the spectacle of machines that were “mechanical and divine.”⁵² Their performances were knowing ones, inspiring as much humor as devotion, but nevertheless the magic was there in works such as the Rood of Grace at Boxley Abbey, which greeted pilgrims annually in the sixteenth century with a Christ who moved and rolled his eyes. Human craftsmen and performers were essential to these performances, simulating responsiveness and awareness to effectively engage the crowds. Something of this knowing spectacle remained in Kasparov’s matches with Deep Blue, including his persistent efforts to deploy openings and play styles that would throw off IBM’s algorithms.⁵³ The human engineer was never entirely hidden behind the mechanism, as IBM employees tweaked the system between every game. Perhaps the best example of this relationship was a highly scrutinized move near the end of game one, the truth of which was only revealed in 2012 when IBM researcher Murray Campbell was interviewed by the popular statistician Nate Silver. The move in question had sent Kasparov “into a tizzy” as it seemed to reflect the ambiguity and refinement of a human-level intelligence, and many have suggested it threw off the grandmaster’s concentration for the second game, which he proceeded to lose.⁵⁴ In fact, as Campbell revealed, the move had been a bug, one the engineers corrected after the first match.⁵⁵ Kasparov himself had made magic out of the algorithm, inventing a sophisticated cultural explanation for what was in the end a random computational artifact.

最初的“土耳其人”以及与卡斯帕罗夫的精彩对决将图灵测试投射到了棋盘上，提出了一个存在主义问题：我们是否能够将智能视为一种游戏风格或美学。“深蓝”不仅击败了卡斯帕罗夫，还改变了整个国际象棋世界。如今，年轻的国际象棋大师们可以与更强大的人工智能对手进行训练，在实施层面，一个新的合作游戏空间也应运而生，由人类和机器组成的“半人马”团队持续击败国际象棋大师和纯算法智能。56^最有效的美学是增强，用计算深度补充人类直觉。这种类比也适用于其他实施领域。优步不仅仅是另一种回家的方式；它正在颠覆雇佣交通工具的监管和文化空间，改变我们解读街道的语法和词汇。算法对工作的再造正在改变文化的字面和想象地理，我们将在本卷末尾再次讨论这个主题。

The original Turk and the great match with Kasparov projected the Turing test onto a chessboard, asking the existential question of whether we would be able to recognize intelligence as a kind of style or aesthetic of play. Deep Blue did not merely beat Kasparov; it has transformed the entire chess world. Now young grandmasters can train against superior AI opponents, and a new space of collaborative play has emerged in the implementation gap where “centaurs” comprising human and machine teams consistently beat both grandmasters and pure algorithmic intelligence.⁵⁶ The most effective aesthetic is one of augmentation, complementing human intuition with computational depth. The analogy holds in other areas of implementation as well. Uber is not simply another way to get home; it is upending the regulatory and cultural space of hired transportation, changing the grammar and vocabulary we all use to read the street. The algorithmic reinvention of work is transforming the literal and imaginary geographies of culture, a theme we will return to at the end of this volume.

从1997年深蓝的胜利到亚马逊的Mechanical Turk，这标志着计算工具的逆转：如今，人脑成了黑匣子，被串联成一个由湿件服务器组成的临时网络，用于处理零碎工作。Mechanical Turk既验证了人类思维的独特智能（在各种情境挑战中，其表现仍优于算法智能），又将其置于算法逻辑的框架之下，为与情境无关的工作创建了一个高效的微支付基础设施。专职的“土耳其人”每小时收入约5美元，其中许多人居住在美国（印度是第二大群体）。57^但即使以“工人”一词的典型含义来谈论它，或者说按小时计酬，也是不诚实的，因为该系统的原子化无法提供特定劳动形式的稳定性、规律性或持久性，导致完成特定任务所需的时间差异巨大。就像云端的许多计算机服务器处于闲置状态（2012 年，麦肯锡公司估计，按功耗衡量，闲置率高达 90%）一样，这些工作者中的大多数人大多数时间都没有参与工作。58 ^Mechanical Turk 将界面公司的逻辑与服务器场的算法逻辑融合在一起，界面公司试图激活常备储备，通过将未充分利用的资源商业化，使我们的生活更加经济高效。

The transition from Deep Blue’s victory in 1997 to Amazon’s Mechanical Turk marks the reversal of the computational instrument: now human brains are the black box, strung together into an ad-hoc network of wetware servers for odd jobs. Mechanical Turk both validates the distinctive intelligence of the human mind (which can still outperform algorithmic intelligence on a wide variety of contextual challenges) and subjects it to algorithmic logic, creating an infrastructure of hyperefficient micropayments for context-free work. Dedicated “turkers” earn about $5 an hour, and many of them reside in the United States (with India making up the second largest contingent).⁵⁷ But it’s disingenuous even to speak of workers in the typical meaning of that term, or an hourly wage, since the system’s atomization offers no stability, regularity, or persistence of particular forms of labor, leading to huge variance in the amount of time it takes to complete a particular task. Just as many computer servers in the cloud sit idle (in 2012, McKinsey & Company estimated the percentage as 90 percent when measured by power consumption), most of these workers are not engaged most of the time.⁵⁸ Mechanical Turk melds the logic of interface companies that seek to activate a standing reserve, making our lives more economically efficient by commercializing underutilized resources, with the algorithmic logic of a server farm.

分布式并行计算的算法逻辑意味着市场根本不是“工人”的市场，而是精心划分的人类注意力和能力的市场：没有人会雇佣“Mechanical Turk”；相反，他们购买的是一定数量的重复性任务（例如，10,000 名 HIT 查看特定图像并用文字描述它）。这是计件工作，导致一些研究人员认为该系统将计算机变成了 21 世纪的缝纫机。59^这种说法并不诚实，因为一台配有训练有素操作员的缝纫机可以出租给出价最高的人，但 Mechanical Turk 市场旨在消除工人的几乎所有专业知识或专业化，从而消除任何真正的议价能力。更重要的是，个人电脑并非系统中最关键的技术：界面——Mechanical Turk 平台和与之相连的人脑——才是关键部分。人类注意力的循环构成了Mechanical Turk的核心商品，用户的机器、亚马逊云以及购买这些循环的数据处理器构成了一个完整的计算系统。在这里，Mechanical Turk的文化机器及其有限的计算作用清晰可见：计算机代码分配任务并处理付款，但真正的工作发生在生物处理器中。

That algorithmic logic of distributed, parallel computing means that the marketplace is not one for “workers” at all but for carefully delimited slices of human attention and ability: nobody hires a “Mechanical Turk”; rather, they purchase the performance of some number of repetitive tasks (e.g., 10,000 HITs looking at a particular image and describing it in words). It is piece-rate work, leading some researchers to argue that the system makes the computer the sewing machine of the twenty-first century.⁵⁹ This is disingenuous because a sewing machine with a trained operator can be hired out to the highest bidder, but the Mechanical Turk marketplace is designed to eliminate almost any kind of expertise or specialization among workers, and thereby any real bargaining power. More important, the personal computer is not the most crucial technology in the system: once again the interface—the Mechanical Turk platform and the human brains connected to it—is the essential piece. Cycles of human attention make up the core commodity of Mechanical Turk, with the users’ machines, the Amazon cloud, and the data processors who purchase these cycles comprising a complete system of computation. Here the culture machine of Mechanical Turk and the limited role of computation within it are clearly visible: computer code assigns jobs and handles payments, but the real work is happening in biological processors.

正如界面经济能消除歧义，将其抽象性隐藏在计算的表象之下，Mechanical Turk 就像一张细密的网，能筛选出歧义。以下是一位博主对成为一名土耳其土耳其人的经历的反思：

Just as the interface economy whisks away ambiguity, concealing its abstractions behind the facade of computation, Mechanical Turk is a kind of fine mesh for sifting through ambiguity. Here is the meditation of one blogger on the experience of being a turker:

[典型的工作] 是你的转录 HIT、你的破译这糟糕的笔迹 HIT、你的调查 HIT。在这些任务中，将人类的模糊性转化为计算机风格的二进制是最令人沮丧的，因为你很可能会得出“我不知道”的答案，然后被迫做出选择。我记得一些愤怒地敲击键盘的例子，比如有一次我被要求根据一张黑白照片来判断珠宝是金的还是银的。根本分辨不出来！或者被问到一个人难以辨认的涂鸦看起来更像“X”还是“Y”。两者都不是，看起来像“Z！” ⁶⁰

[The typical jobs] are your transcription HITs, your decipher-this-horrible-handwriting HITs, your survey HITs. These are the tasks where translating human ambiguity into computer-style binary is the most frustrating, because you’re likely to arrive at the answer “I don’t know” and be forced to choose anyway. I remember a few angry-keyboard-mashing examples, like the time I was told to identify whether jewelry was gold or silver based on a single black-and-white picture. Impossible to tell! Or being asked if a person’s illegible scribbling looks more like “X” or “Y.” Neither, it looks like “Z!”⁶⁰

Mechanical Turk 是另一个在实施差距中运作的套利系统。事实上，它将这一差距量化并商品化，将其转化为一系列微任务和判断：抽象和具体化的增量时刻。它将这些枯燥乏味的相同任务外包出去，然后像全动态电影中的单个帧一样，被整合成一种连续计算的幻象。Mechanical Turk 的独特之处在于它将人类计算的后端以商业服务的形式展示出来，将界面经济的逻辑应用于实施区域本身。正如上文所述，在这个充满争议的空间中运作的人类必须不断地在计算和文化的意义机制之间进行协商。

Mechanical Turk is another system of arbitrage operating in the implementation gap. In fact, it quantifies and commoditizes that gap, turning it into a series of micro-tasks and judgments: incremental moments of abstraction and concretization. The grinding series of identical tasks it farms out can then be integrated, like individual frames in a full-motion film, into an illusion of continuous computation. What makes Mechanical Turk unusual is the way it puts the human back end of computation on display as a commercial service, applying the logic of the interface economy to the zone of implementation itself. As the quote above demonstrates, the humans who operate in that contested space must constantly negotiate between computational and cultural regimes of meaning.

计件诗学

Piecework Poetics

在我们列举的各种例子中，界面层不仅改变了政治，也改变了劳动的美学，将新的算法语境强加于身份、价值和成功等核心问题。这场巨变已经彻底颠覆了出租车司机和合同工的生活，也影响着医生和律师。我们许多人随身携带的智能手机正在不可逆转地模糊几乎每个全球界面经济参与者的职业角色和界限，以我们才刚刚开始理解的方式重新定义工作的影响。

Across our various examples the interface layer is transforming not only the politics but the aesthetics of labor, imposing new, algorithmic contexts on central questions of identity, value, and success. The sea change that has already upended the lives of cab drivers and contract workers is also affecting doctors and lawyers. The smartphones so many of us carry are irrevocably blurring professional roles and boundaries for almost every participant in the global interface economy, redefining the affect of work in ways that we are only beginning to understand.

Mechanical Turk 是这些转变的熔炉，它是一个专门为商业化“人类是计算机器的重要齿轮”这一论点而创建的人工市场。处于这一系统核心的土耳其人不仅承担着管理模糊性的无尽微任务的挑战，他们还承担着在计算应用程序中充当人类元素的情感工作。概念艺术家尼克·瑟斯顿在《分包合同或诗意权利原则》中挖掘了这一实施差距，这是一部主要由土耳其人自己委托创作的对 Mechanical Turk 的文学探索作品。《分包合同》巧妙地改变了平台的核心前提，要求其湿件服务器承担一项任务，即诗歌创作，而在这一任务中，人类仍然比算法占据绝对优势。

Mechanical Turk is a crucible for these transitions, an artificial market created specifically to commercialize the thesis that humans are important cogs in computational machines. The turkers at the heart of this system not only take on the challenge of endless micro-tasks managing ambiguity—they also take on the affective work of acting as a human element inside of a computational application. Conceptual artist Nick Thurston mined this seam of the implementation gap in Of the Subcontract, Or Principles of Poetic Right, a literary exploration of Mechanical Turk largely commissioned from turkers themselves. Of the Subcontract neatly inflects the platform’s central premise, asking its wetware servers to take on a task, poetic composition, where humans still hold a decided edge over algorithms.

这本诗集的诗歌之所以如此震撼，正是因为它们如同机器一般。每首诗的开头都会列出创作时间（通常以分钟和秒为单位）、创作者的时薪，以及收录在书中的HIT或投稿（例如，七篇投稿中的第四篇，两篇投稿中的第一篇）。整本诗集的序言名义上是由麦肯齐·沃克撰写的，但实际上是一位拉合尔的自由撰稿人代笔，正如后记所示。

The poems in the collection are powerful precisely because of their pseudo-automaton status. Each piece begins with a readout of the amount of time spent on composition (usually measured in a few minutes and seconds), the hourly rate of this particular worker, and an indication of which HIT or submission was printed in the book (e.g., the fourth of seven submissions, the first of two). The entire collection is prefaced by an introduction nominally authored by McKenzie Wark but actually ghostwritten by a Lahore-based freelancer, as the afterword reveals.

这本诗集极具美学意义，因为它既吸纳了计算时间的“暴政”，又对其进行了质疑。Turkers——那些以平均每小时5美元的价格争夺HIT的人——可能并不觉得自己有能力创作诗歌。使用Turk系统将创造力分配为微任务，营造出一种独特的情感荒诞感。

The collection is a particularly powerful aesthetic move because it simultaneously co-opts and questions the tyranny of computational time. Turkers—people competing for HITs at an average of $5 an hour—are probably not people who feel like they can afford to compose poetry. Using the Turk system to assign creativity as micro-task performs a particular kind of affective absurdity.

这是人工文学。首先，这些人被要求通过亚马逊界面模仿他们的人性。其次，这些诗歌并非出于个人创作冲动而创作，而是外部强制的计件工作——一个虚拟的诗歌工厂。印刷页面的格式，以及随之而来的关于成本和创作时间的元数据，强调了双重人工性，即使诗歌本身依然存在：无论其高度商品化的创作条件如何，它们都是人类的写作作品，传达着特定的人类思想。

This is artificial artificial literature. First, these humans are asked to emulate their humanity through the Amazon interface. Second, these are not poems authored through an individual impulse to create but as externally mandated piecework—a virtual factory for verse. The formatting of the printed page, with its attendant metadata about cost and composition time, emphasizes that double artificiality even as the poems themselves remain: no matter the conditions of their hyper-commoditized composition, they are works of writing by humans, conveying particular human ideas.

这本书的第一部分诗歌名为“人工人工智能”，不断闪现出潜伏在系统之下或背后的人性，比如第 0.24 首“我的儿子”：“你是我胖乎乎的泡泡小可爱 / 我爱你我的儿子 / 因为你是唯一。” ⁶¹要求这些土耳其机器人将自己写入系统界面，立刻就将其他人也带入了视野。我们瞥见了儿女、失踪的父母和失去的爱人，所有这些都经过了 Mechanical Turk 的匿名机制及其泰勒式的劳动和价值计算。这些正是界面层通常会消除的语境联系，它们会通过诗人的声音悄悄地回来。

The book’s first section of poems, titled “Artificial Artificial Intelligence,” generates persistent flashes of the humanity lurking beneath or behind the system, like entry 0.24, “My Son”: “You are my chubby bubble cutie pie / I love you my son / For you are the only one.”⁶¹ Asking these turkers to write themselves into the interface of the system immediately brings other humans into view along with them. We catch glimpses of sons and daughters, missing parents and lost loves, all filtered through the anonymizing mechanism of Mechanical Turk and its Taylorist computations of labor and value. These are exactly the kinds of contextual links that the interface layer generally eliminates, creeping back in through the voice of the poet.

许多诗歌似乎像 Mechanical Turk 系统一样，用平淡无味的语调回应着提示，提供一些陈词滥调和老生常谈的平淡内容，只需 12 秒即可完成创作。但对另一些诗歌而言，一种谨慎的启示感和情感的流露，将按分计酬的诗歌创作转化为美的表达。除了默默地展现人性之外，这些诗人还在算法系统中持续不断地探索身份认同的问题，例如以下节选的 0.04 号作品：

Many poems seem to respond to their prompts with the same flat, affectless tone as the Mechanical Turk system itself, offering up anodyne confections of cliché and truism, completing the task of composition in as little as twelve seconds. But for others a cautious sense of revelation, of emotional exposure, transforms the grim performance of poetry-by-the-cent into something beautiful. Beyond quietly asserting their humanity, these poets also persistently grapple with the question of identity in an algorithmic system, like entry 0.04, excerpted here:

你把我放在一个有锁和钥匙的盒子里。

我时不时地撬锁。

但获得自由却并不是那么简单。

我被你赋予我的头衔束缚住了。

我多么渴望做我自己，多么渴望挣脱束缚。

但我的恐惧让我留在那个地方。

那个你强迫我去的地方。

我最大的恐惧是你不爱我。

我是我所认为的那个人吗？

或者我是你所说的那个人？⁶²

You have put me in a box with a lock and a key.

I pick at the lock from time to time.

But it is not so simple to be free.

I am chained by the title you have put on me.

I so long to be me, to break loose.

But my fears keep me in that place.

That place where you forced me to be.

My greatest fear is of you not loving me.

Am I who I think I am?

Or am I who you say I am?⁶²

在八分三十秒的时间里，作者探讨了身份的不确定性，其视角似乎同样适用于爱人和亚马逊本身，他将强加的身份和头衔的锁描述为一种不仅阻碍运动而且阻碍确定性本身的钳子。

In eight minutes and thirty seconds the author explores the uncertainty of identity in terms that seem to apply equally to a lover and to Amazon itself, describing the lock of imposed identity and title as a vise that thwarts not just movement but certainty itself.

时间戳、货币印记以及作者的匿名性，都表明《分包合同》的地位如同另一种图灵测试。这些被分包的作者至少在两个层面上运作。在第一个层面上，他们在滴答作响的时钟和等待的HTML表单字段的约束下，努力展现人性的形象。但在第二个层面上，他们仍然像计算系统中的齿轮一样运转，多次重复同一项任务，并试图优化自身的处理效率，以尽快完成每一个诗意的“命中”。然而，在这些抽象的、人工的层层叠加之下，美的瞬间浮现。

The time stamp, the money stamp, and the anonymity of the authors all convey Of the Subcontract’s status as another kind of Turing test. These subcontracted authors are performing on at least two levels. On the first level, they are pulling off an impression of humanity under the constraints of a ticking clock and waiting HTML form field. But on the second, they are still performing as cogs within a computational system, repeating the same task multiple times and attempting to optimize their own processing efficiency to complete each poetic HIT as quickly as possible. And yet, beneath all these layers of abstracted, artificial production, moments of beauty emerge.

《转包合同》是一场融合超现实主义和威权主义的文学实验——就好像有人收集了特里吉列姆黑暗奇幻电影《妙想天开》中居民的一本诗集。它与超现实主义思维以及自动写作、绘画和其他创作过程有着共同的根源，但截然不同的是，这里不是创意精英的游乐场，而是社会经济弱势群体的工厂。创意实验的乐趣和责任在于《转包合同》的真正作者尼克瑟斯顿和他的快乐团队，他们委托创作诗歌，这些诗歌起初是他们的小游戏，最终成为为他人匿名创作的作品。这种紧张和不适正是瑟斯顿和他的合作者想要达到的效果。

Of the Subcontract is a literary experiment uniting surrealism and authoritarianism—as if someone had collected a volume of poems from the denizens of Terry Gilliam’s darkly fantastical film Brazil. It shares roots with surrealist thinking and automatic writing, drawing, and other creative processes, with the sharp difference that this is not a playground for creative elites but a factory for the socioeconomically disadvantaged. The pleasure and responsibility of the creative experiment resides with Of the Subcontract’s real authors, Nick Thurston and his merry band, who are commissioning poetry that began as their little game and ends as work for anonymous others. This tension and discomfort are precisely what Thurston and his collaborators sought to achieve.

在如此尖锐愤世嫉俗的艺术实验的夹缝中，美学的复兴或许正是他们所期盼的：在诗歌破产的证据边缘，证明其价值。《分包合同》赋予美学其在其他价值形式交汇处的传统地位，使金融交易、算法计算和人工计算得以转化为拥有艺术生产官方印章的东西。众包计件诗歌的政治舞台也变成了一种性质不同的行为艺术，一种不​​仅维护批评本身价值，也维护批评所产生的过程性副产品价值的文学作品。我们最终欣赏的不仅是这本书的艺术理念，还有计件诗歌本身。通过这种方式，《分包合同》对“土耳其机器人”（Mechanical Turk）进行了算法式的批判，依靠系统本身来处理或运行这种批判。因此，作为读者，我们的工作就是审视委托创作和汇编这些诗歌的迭代过程。这些匿名 turker 作品的收集和语境化的实施方式是整个诗歌机制的一个组成部分。

The resurgence of aesthetics between the cracks of such a pointedly cynical artistic experiment is presumably exactly what they were hoping for: a proof of the value of poetry written in the margins of a proof of its bankruptcy. Of the Subcontract accords aesthetics its traditional place at the intersection of other forms of value, allowing financial transactions, algorithmic calculations, and human computation to be transmuted into something with the official seal of artistic production. The political theater of crowdsourcing piecework poetry also becomes performance art of a different nature, a literary production that asserts the value not only of the critique but of the processual by-products generated by the critique. We end up admiring not just the artistic concept of the volume but the piecework poems themselves. In this way, Of the Subcontract performs an algorithmic critique of Mechanical Turk, relying on the system itself to to process or run that critique. Our work as readers, then, is to examine the iterative process of commissioning and assembling these poems. The implementation, the ways that the work of these anonymous turkers was collected and contextualized, is an integral part of the whole poetic mechanism.

实验诗人达伦·沃什勒 (Darren Wershler) 在他的后记中触及了这一批判性纠葛的核心，值得大段引用：

Experimental poet Darren Wershler strikes at the heart of this critical tangle in his afterword, which is worth quoting at length:

我们也读过一些文章，解释“土耳其人”实际上是全球化和网络化环境下工人不稳定状况的一个优雅隐喻。我们也创作了大量利用亚马逊 Mechanical Turk 作为生产媒介的艺术作品，来阐述同样的观点，但方式更艺术化。

重点不在于这个机制是空的，就像某种中立的复制器。重点在于，它本身就包含了一个为你设定的位置——就像弗朗茨·卡夫卡小说《审判》中的法则——无论这个位置是作为玩家在它面前，作为操作者在它内部，作为观众在它背后被展示其误导性组成部分，作为评论家在远处借助你的批评来描述和揭开它的神秘面纱，或者，越来越多地，作为艺术家或作家在你的项目中（误）使用它。当你把这个设定作为一个问题来处理时，机器就会开始运转^。63

We have also read essays explaining that the Turk is in fact an elegant metaphor for the precarious condition of the worker in a globalized and networked milieu. And we have made a substantial amount of art that actually makes use of Amazon Mechanical Turk as a productive medium to demonstrate the same point, but in a way that is, you know, artier.

The point is not that the mechanism is empty, like some kind of neutral reproducer. The point is that it is a mechanism that already includes a spot for you—like the Law in Franz Kafka’s novel The Trial—whether that spot is in front of it as a player, inside it as the operator, behind it as the spectator being shown its misleading components, from afar as the critic describing and demystifying it by virtue of your criticism, or, increasingly, as the artist or writer (mis)using it in your project. The moment that you engage the setup as a problematic the machine springs into action.⁶³

Wershler 的论点完美地捕捉了算法在更广泛的劳动和文化生产语境中的现实：系统不仅重构了生产方法和产出，还重塑了整个文化框架，将我们所有人都卷入其中，成为合作者。正如 Galloway 在《界面效应》中所说：

Wershler’s argument beautifully captures the reality of algorithms in the context of labor and cultural production more broadly: the system reconfigures not just the methods and outputs of production but the entire cultural frame, implicating all of us as collaborators. As Galloway puts it in The Interface Effect:

我反对那种意识形态上的迷思：我们自由，而中国孩子却被锁链束缚；我们的电脑是生命线，而他们的电脑是祸害。这种污言秽语必须被摒弃。我们都是“打金农”，而且更矛盾的是，我们大多数人都是心甘情愿地这样做，而且根本不赚钱^。64

I dispute the ideological mystification that says that we are the free while the Chinese children are in chains, that our computers are a lifeline and their computers are a curse. This kind of obscenity must be thrown out. We are all gold farmers, and all the more paradoxical since most of us do it willingly and for no money at all.⁶⁴

我们每个人都在生活中“误用”这些系统，无论我们将其视为文化或审美实践、私人或职业交流，还是对这些系统的批判。人类与算法文化机器之间密不可分的混合劳动，无时无刻不在我们身边发生着。从搜索栏到微软Word，我们以最平庸、最普遍的方式依赖算法系统来执行和评估劳动。归根结底，《转包合同》的核心叙事并非受压迫的土耳其工人对其人类处境的短暂描述，而是我们每个人在计算鸿沟中所处的位置。

We are all “(mis)using” these systems in our lives, whether we consider that use cultural or aesthetic practice, private or professional communication, or even the critique of these systems. Hybrid labor between inextricably linked human and algorithmic culture machines is happening all around us, all the time. In the most banal, pervasive ways, we rely on algorithmic systems to perform and evaluate labor, from the search bar to Microsoft Word. In the final analysis, Of the Subcontract’s central narrative is not that of oppressed turkers articulating brief glimpses of their human condition, but of the place held for each of us in the computational gap.

道德机制

Moral Machinery

关于机械化伦理影响的争论由来已久。例如，在工业化鼎盛时期，最初的国际象棋土耳其人形象在美国长期公开亮相，激发了华丽的文笔和全神贯注的观众。65^如今，当我们思考界面层和新的行为修改算法系统时，这些争论的基调仍然具有启发意义，这些系统远比19世纪工厂的残酷条件更为客观。即使在当时，许多人也认为自动化的兴起是一种道德良善的力量，激励（或迫使）工人以与他们所操作的机器相同的奉献精神行事。这就是一位英国经济学家在1835年所说的“道德机器”：一套管理干预系统，旨在增强工业体系对人类工人秩序和生产力的自然倾向^。66

Debates about the ethical impact of mechanization are anything but novel. The original chess-playing Turk, for example, inspired florid prose and rapt audiences during its long run of public appearances in the United States in the heyday of industrialization.⁶⁵ The tenor of those debates is instructive today when we think of the interface layer and new algorithmic systems of behavior modification that are far less blunt than the brutal conditions of nineteenth-century mills and factories. Even then, many viewed the rise of automation as a force for ethical good, inspiring (or forcing) workers to conduct themselves with the same dedication as the machines they attended. This is what one British economist called “moral machinery” in 1835: a system of managerial interventions to enhance the industrial system’s natural tendencies toward order and productivity among human workers.⁶⁶

正如历史学家斯蒂芬·P·赖斯 (Stephen P. Rice) 所说，土耳其机器人的奇观为美国观众塑造了一种新的影响，它所进行的劳动模糊了机器与人类、算法与工人之间的界限。

As historian Stephen P. Rice argues, the spectacle of the Mechanical Turk modeled a new affect for its American audiences, performing a form of labor that blurred the line between machine and human, between algorithm and worker.

在中产阶级对工人自我控制能力焦虑的背景下，棋手扮演着双重角色：受监管的机器和理想的机械化工人。观众可以在棋手身上找到人类独有的意志和理性特质，而不必将这些特质与规律性和效率等机械特质区分开来。[土耳其人]被解读为一台受监管或“有意识的”机器，展现了新的生产秩序^。67

Launched into the scene of middle-class anxiety about worker self-control, the chess-player assumed the twin statuses of regulated machine and ideal mechanized worker. Viewers could locate in the chess-player the uniquely human traits of will and reason without having to remove those qualities too far from the mechanical traits of regularity and efficiency. Read as a regulated or “minded” machine, [the Turk] showed the new productive order in place.⁶⁷

这种人机双重框架与亚马逊土耳其机器人（Amazon Mechanical Turk）的界面如出一辙：每个“任务完成”（HIT）都承诺给任务负责人一个既可靠又可靠的响应。《分包合同》（Of the Subcontract）的诗歌将这种二元论推向极致，在机器内部开辟了一条诗意的自我反思之路。

This double framing as human and machine closely echoes the interface of Amazon Mechanical Turk, where each HIT promises the taskmaster a response that is both mechanically reliable and reliably human. Of the Subcontract’s poems push that dualism to its limit, opening up a channel for poetic self-reflection from the very belly of the machine.

19 世纪和 21 世纪的土耳其平台都运用美学来为用户和观众营造特定的态度或情感。两台机器都从事着特定的工作（例如下棋、完成数百万次分布式 HIT），但它们也承担着另一种劳动，强化了赖斯所指的人机二元论。这就是政治哲学家迈克尔·哈特和安东尼奥·内格里所说的“情感劳动”，即产生物质和情感产出的经济劳动，经典的例子是零售业的“微笑服务”。68对于¹⁹世纪的土耳其人来说，奇观本身就是情感的产物：土耳其人挥舞着手臂的身影，摆满无用（却令人印象深刻）齿轮和机械的展示柜，当然还有隐藏在机器秘密隔间内的人类的劳动，所有这些都有助于制造出一种惊叹和兴奋的感觉。在21世纪，亚马逊的Mechanical Turk推广了重复性和可替代性的概念，暗中鼓励其土耳其人以及人工计算的购买者将每个工人视为一个简单的数字。HIT本身是按批订购的，正如我们上文所述，它们通常在几秒钟内消失，这鼓励土耳其人时刻警惕新的机会。通过依赖亚马逊的任务分配网络和个人电脑，该系统还限制了用户的身体和时间（取决于任何特定时间可用的计件工作种类的波动以及这些工作的报酬情况）。该系统不仅生产HIT，也围绕这些HIT形成了一种劳动文化。

The nineteenth- and twenty-first-century Turk platforms both use aesthetics to produce certain attitudes or affects for their users and viewers. Both machines do a particular kind of work (playing chess, completing millions of distributed HITs), but they also perform another kind of labor when they reinforce the dualism between human and machine that Rice identifies. This is what political philosophers Michael Hardt and Antonio Negri call “affective labor,” or economic work that produces a physical and emotional output, classically exemplified by the retail dictate “service with a smile.”⁶⁸ For the nineteenth-century Turk, the spectacle itself is the affective product: the figure of the Turk with his moving arm, the display cabinet full of functionless (but impressive) gears and machinery, and of course the labor of the man hidden inside the machine’s secret compartment all contribute to manufacturing a sense of wonder and excitement. In the twenty-first century, Amazon’s Mechanical Turk promotes notions of repetition and fungibility, implicitly encouraging its turkers as well as the purchasers of human computation to think of each worker as a simple number. The HITs themselves are ordered up in batches and, as we saw above, often disappear in seconds, encouraging turkers to be constantly vigilant for new opportunities. By relying on Amazon’s task distribution network and personal computers, the system also constrains the bodies and the time of its users (depending on the fluctuations of what piecework is available at any given time, and how well that work is compensated). The system produces HITs but also a kind of labor culture around those HITs.

这种劳动文化的痕迹也体现在共享经济的应用程序和界面中。正如Uber和Lyft的广告所暗示的那样（图4.3、图4.4），界面经济中大量的精力都投入到消费者和服务提供者之间的情感生产中。用户反馈、情感享受以及新自由主义的独立理想，定义了这些系统的各个层面，从它们的标识到反馈机制。它们是极具说服力的平台，旨在创建一个由算法介导的社群空间。用户和提供者之间的计算层负责管理细节，计算反馈分数，使特定用户彼此可见或不可见，并总体上管理着使用系统的“体验”。这些服务以透明度为基础，声称用户反馈和背景调查的结合能够以比我们通常自行审查更严格的方式审查我们的司机、清洁工、跑腿人员等等。但系统仍然控制着这种透明度的部署，选择我们看到的信息，并在算法黑匣子背后控制着这个验证过程。最后，我们通常会看到少数几个强烈推荐的选项：精心设计的用户体验的第一步，它既取决于情感结果，也取决于财务结果。

That trace of a labor culture also appears in the apps and interfaces of the sharing economy. As the ads for Uber and Lyft suggest (figure 4.3, figure 4.4), a huge amount of energy in the interface economy goes toward the production of affect among consumers and providers of services. User feedback, emotional enjoyment, and a neoliberal ideal of independence define these systems at every level, from their logos to their feedback mechanisms. They are persuasive platforms designed to create an algorithmically mediated space of community. The computational layer between users and providers manages the details, calculating feedback scores, making particular users visible or invisible to one another, and generally managing the “experience” of using the system. These services build their reputation on transparency, arguing that the combination of user feedback and background checks serves to vet our drivers, house cleaners, errand-runners, and so forth in a more robust way than we would typically do on our own. But the systems retain control over the deployment of that transparency, choosing what information we see and maintaining control of this validation process behind algorithmic black boxes. In the end what we typically see are a small number of options that are all strongly recommended: the first step in a carefully choreographed user experience that depends as much on emotional outcomes as financial ones.

金融交易的情感价超越了情感劳动，其根源可以追溯到亚当·斯密的《道德情操论》 ，这与他在《国富论》中关于资本主义的著名论点形成了重要的对比。斯密在《道德情操论》中指出，社会凝聚力和市场的有效运作依赖于以想象力为根基的道德行为逻辑。我们永远无法了解他人的生活经历，但社会参与依赖于想象这种经历的快乐和痛苦。同理心是所有社会交往的重要组成部分，是一种根据道德准则和我们自己的判断调整自身行为的反馈机制。斯密认识到我们永远无法真正了解他人的经历，但却不断努力去想象，这给每个文化实践参与者带来了重要的道德责任，这与他的自由主义政治观点相呼应。他认为，道德社会的基础是我们所有人不断进行富有想象力的工作，以描绘出同理心的领域，创造一种务实的正义感和指导经济和社会行为的共同经验。

The emotional valence of financial transactions goes beyond affective labor, tracing its roots to Adam Smith and the Theory of Moral Sentiments, an important counterpoint to his better-known arguments about capitalism in The Wealth of Nations. Smith argues in Moral Sentiments that social cohesion and the effective functioning of a marketplace depend on a logic of virtuous action that has imagination at its root. We can never know the lived experience of others, but social engagement depends on imagining that experience in its joy and pain. Empathy is a crucial component of all social intercourse, a feedback mechanism for adjusting our own behavior according to moral guideposts and our own judgments. Smith’s recognition that we can never truly know the experience of others—yet constantly strive to imagine it—places an important moral responsibility on each participant in cultural practice, one that parallels his libertarian political views. The foundation of a moral society, he argues, is the constant imaginative work we all do to map out the territory of empathy, creating a pragmatic sense of justice and shared experience that guides economic and social behavior.

正如法学教授杰迪戴亚·珀迪 (Jedediah Purdy) 所言，将特定的情感甚至身份地位经济强加于服务人员，为史密斯的模型带来了一些独特的新意：“强制性微笑是资本主义核心讽刺的一部分……假装微笑是新的封建主义。它是情感劳动经济秩序的关键，这种秩序告诉人们，根据他们在社会和经济等级中所处的位置，他们是谁以及如何成为他们。” ⁶⁹当我们将道德情操理论应用于算法时代时，我们可以看到第三方进入了人类行为者之间的交易空间。界面经济的扩张及其对情感劳动的依赖为新的表演类型带来了生机：我们现在是界面的重要组成部分，用我们的身体、思想和情感为计算的美学上层建筑做出贡献。到处都是五星好评！然而，我们也是作为“半人马”进行这项工作的，依赖于相同的界面平台来获取基本的情感线索。将各种金融和非商业安排转变为可雇佣服务的“Uber for X”系统的进程带来了服务经济的情感劳动，但计算执行了同理心富有想象力的工作的关键部分。

As law professor Jedediah Purdy argues, the economic imposition of particular emotions and even identity positions on service workers brings something distinctively new to Smith’s model: “Mandatory smiles are part of an irony at the heart of capitalism. … Faking it is the new feudalism. It is the key to an economic order of emotional work that tells people who and how to be on the basis of where they fall in the social and economic hierarchy.”⁶⁹ When we adapt the theory of moral sentiments to the age of the algorithm, we can see that a third party enters the transactional space between human actors. The expansion of the interface economy and its dependence on affective labor brings new genres of performance to life: we are very much part of the interface now, contributing to the aesthetic superstructure of computation with our bodies, our minds, and our affect. Five star ratings all around! Yet we do this work as “centaurs” too, depending on the same interface platforms for essential affective cues. The march for “Uber for X” systems that turn various financial and noncommercial arrangements into services for hire brings with it the affective labor of the service economy, but with computation performing crucial pieces of the imaginative work of empathy.

以优步为例。该公司的反馈系统要求乘客和司机互相评价，从而创建一个持久的用户档案数据库，使公司能够识别最顺从和最麻烦的员工和客户（然后可以主动或被动地将他们排除在未来的交易之外）。这种消费者反馈的概念依赖于史密斯所定义的同理心的资本主义功能：创建一个同理心框架，一个反馈渠道，可以促进更好的互动。像知道司机的名字这样简单的事情，比我们大多数人与出租车司机的沟通更能激发同理心。当然，这些数据可以帮助公司识别其商业实践中的问题，并帮助客户找到他们想要的服务和产品。反馈数据使我们能够超越个人体验进行推断，汇总同理心想象的行为，从而构建更全面的未来视角（例如，43 个人认为这位司机很粗鲁）。但是，通过创建推断工具，这些系统也用一系列调查问卷取代了史密斯所设想的想象行为的细微差别和作者身份。将同理心量化，通常用一个完全抽象的五星级量表来衡量，这鼓励我们将繁重的工作留给算法。我们不再自己动手，而是将其外包给一台文化机器，它由定制的李克特量表问题和评论表单、黑箱排名算法，以及由算法中介的陌生人群体构成，而界面公司正试图将这些人整合成一个虚拟社区。这种做法的变体已经持续了几个世纪，从“五分之四的医生选择产品 X”这样的统计数据，到广告中普遍存在的同理心诉求——但计算平台将其提升到了更深的认识论层面。

Take Uber, for example. The company’s feedback system asks both riders and drivers to rate one another, creating a persistent database of user profiles that allows the company to identify its most compliant and troublesome workers and clients (who can then be passively or actively excluded from future transactions). This notion of consumer feedback depends on the capitalistic function of empathy as Smith defined it: creating an empathetic frame, a channel for feedback, encourages better interactions. Something as simple as knowing your driver’s name is a more empathetic contact than most of us have with a taxi driver. And of course this data can help companies identify problems in their business practices and help customers find the services and products they want. Feedback data allows us to extrapolate beyond the individual experience, aggregating acts of empathetic imagination to project a more holistic view of the future (e.g., forty-three people thought this driver was rude). But by creating the tool for extrapolation, these systems also replace the nuance and authorship of the imaginative act as Smith imagined it with a series of surveys. Quantifying empathy, often into a completely abstracted five-star scale, encourages us to leave the heavy lifting to the algorithms. Instead of doing the imaginative work ourselves, we outsource it to a culture machine built up of tailored Likert-scale questions and comment forms, black box ranking algorithms, and the algorithmically mediated crowd of strangers that interface companies seek to synthesize into a virtual community. Variations on this move have persisted for centuries, from the deployment of statistics like “four out of five doctors choose product X” to the empathetic appeals of advertising in general—but computational platforms take it to a deeper epistemological level.

正如我们上文所述，“共享经济”的核心关键词并非想象力，而是信任。Uber及其同行愿意承担各种形式的文化和想象力工作，以换取我们的信任（当然还有我们的金钱）。他们要求员工投入的情感劳动服务于更大的信任，从而打造出一个始终如一、友好服务的品牌。最重要的是，我们被要求信任算法。相信反馈和背景调查会剔除任何危险或粗鲁的司机。相信系统会找到最近、最好的车辆送你上路——相信卡通地图是真实的。相信司机会得到公平的报酬，相信你会支付合理的价格。在所有这些主张中，同理心的想象力工作都被外包给了算法。我们不需要弄清楚如何打车、如何评估车辆及其司机、如何计算小费，或者参与任何其他定义出租车体验的细微行为和共情时刻。它们不会轻易消失。相反，Uber 要求我们用算法的方式去做一些替代性的共情工作。其中一些工作是明确要求或强制的：给司机评分，在社交媒体上分享体验以获得未来的折扣，招募朋友使用该服务以获得更多折扣。还有一些工作则不那么明显：Uber 的广告和员工文化要求我们认可司机，将他们视为创业者、个人主义者和自由职业者。不同的公司对这种认可的呈现方式不同（Lyft 鼓励乘客坐在前排；Uber 则不然），但它们都致力于构建一个由算法维系的合成实践社群。

As we saw above, the central keyword for the “sharing economy” is not imagination, but trust. Uber and its peers offer to take on various forms of cultural and imaginative work in exchange for our trust (and, of course, our money). The affective labor they ask their workers to engage in serves that larger trust, building up a brand of consistent, friendly service. We are asked to trust, most significantly, in the algorithm. Trust that the feedback and background checks will weed out any dangerous or merely rude drivers. Trust that the system will find the closest, best vehicle to whisk you on your way—that the cartoon map is real. Trust that the driver will be fairly compensated and that you will pay a fair price. In each of these claims, the imaginative work of empathy has been outsourced to the algorithm. We don’t need to work out how to catch a cab, assess the vehicle and its driver, calculate the tip, or engage in any of the other micro-practices and empathetic moments that define the taxi experience. They do not simply disappear. Instead, Uber asks us to perform substitute empathetic work, in algorithmic terms. Some of this work is explicitly requested or required: rate the driver, share the experience on social media for a future discount, recruit friends to the service to get another discount. Some of it is less obvious: the ads and the labor culture of the service ask us to validate the drivers as fellow entrepreneurs, individualists, and free agents. Different companies present different versions of that validation (Lyft encourages riders to sit in the front; Uber does not), but they all seek to build a synthetic community of practice held together by algorithms.

这可以催生出强有力的新型社区和社会行动，例如一位优步司机为一位身患绝症的乘客筹款的故事，他部分是通过动员优步社区来实现的。70^但它也通过算法中介改变了经济生活的一个核心方面，将至关重要的共情演算嵌入到文化机器中。我们才刚刚开始看到这一举措的影响，即人们越来越依赖算法本身来执行情感劳动。第二章中Siri的彩蛋提供了一个机器为我们执行情感工作的例子。快速发展的量化健康和福祉领域提供了一个更广阔的舞台，算法系统试图与我们的身体协同工作，并通过我们的身体执行情感工作。有时，我们似乎迫切地寻求直接执行和体验这种情感劳动的渠道，或许是因为其中很大一部分已经被算法、用户界面和计算后端外包和形式化了。否则，为什么用一个允许你给朋友发送拥抱的Facebook应用就能如此轻松地赚取数千美元呢？对情感交流的渴望，对一个能够直接想象的空间的渴望，标志着抽象与实现之间的另一道鸿沟。人类与机器在文化中行动者之间富有想象力的共情之间的鸿沟，归根结底在于价值的建构。正如斯密试图将经济实践置于主体间性、共情性理解的基础之上一样，我们现在正努力在算法世界中定义价值的基本结构。

This can result in powerful new forms of community and social action, like the story of an Uber driver who raised money for a terminally ill passenger, in part by mobilizing the Uber community.⁷⁰ But it also reroutes a central facet of economic life through algorithmic mediation, embedding the essential empathetic calculus in the culture machine. We are just beginning to see the impact of this move, a growing dependence on algorithms themselves to perform affective labor. The Easter eggs in Siri from chapter 2 provide one example of a machine performing emotional work for our benefit. The rapidly expanding field of quantified health and well-being offers a broader arena where algorithmic systems seek to collaboratively perform affective work with and through our bodies. At times it seems like we desperately seek channels to perform and experience this affective labor directly, perhaps because so much of it has been outsourced and formalized by algorithms, user interfaces, and computational back ends. Why else would it be so easy to make thousands of dollars with a Facebook app that allows you to send hugs to your friends? That hunger for emotional contact, for a space where we can imagine directly, marks another disparity between abstraction and implementation. The gulf between the imaginative empathy of human and machine actors in culture comes down to the construction of value. Just as Smith sought to put economic practice on a foundation of intersubjective, empathetic understanding, we are now struggling to define the fundamental structure of value in an algorithmic world.

笔记

Notes

边缘殖民

Colonizing the Margin

2010年5月6日，美国资本市场遭遇暴跌，道琼斯工业平均指数的跌幅超过其110年历史上的任何一天。但几分钟后，市场反弹，主要公司股价在跌至每股几美分后又回归正常价格。整个事件似乎来得快去得也快，留下了一系列关于塑造世界金融市场力量的悬而未决的问题。这种剧烈动荡的根源在于一种高度专业化且利润丰厚的算法套利形式。

On May 6, 2010, the U.S. capital markets experienced a tremendous crash, with the Dow Jones Industrial Average losing more value than on any other day in its entire 110-year history. But within minutes, the market rebounded, and after major company stocks had dropped to pennies a share they returned to normal pricing. The entire episode seemed to fade away as quickly as it appeared, leaving a series of hanging questions about the forces shaping world financial markets. At the root of this radical instability is a highly specialized and lucrative form of algorithmic arbitrage.

在过去十年中，计算系统对市场的重要性日益增加，它逐渐用服务器机柜和专有代码的黑暗世界取代了人们在拥挤的交易大厅里穿着鲜艳背心大喊大叫的熟悉景象。正如财经记者迈克尔·刘易斯在《快闪小子》中所描述的那样，算法的交易作用已经从简单地执行交易发展成为对整个市场体系的一种微税。1^这是一个典型的套利例子，或者说利用不同市场来创造利润，比如交易者在乡下买入廉价小麦，然后在城里卖出赚取利润。归根结底，套利总是与时间有关——在别人为你抓住差价之前——而对于算法来说，时间以微秒为单位。在华尔街，这被称为高频交易 (HFT)。

Over the past ten years, computational systems have become increasingly important to the markets, gradually replacing the familiar sight of men in bright-colored vests yelling across a crowded trading floor with the tenebrous world of server cabinets and proprietary code. As financial journalist Michael Lewis describes in Flash Boys, the transactional role of algorithms has grown from the simple execution of trades into a kind of micro-tax on the entire market system.¹ This is a classic example of arbitrage, or the leveraging of different markets to create profit, like traders who buy cheap wheat in the country to sell at a profit in the city. Ultimately arbitrage is always about time—capturing the difference in price before someone else does it for you—and for algorithms, time is measured in microseconds. On Wall Street, this is now known as high frequency trading (HFT).

^20世纪90年代和21世纪初，证券交易所开始数字化处理交易并蓬勃发展，从纽约证券交易所和纳斯达克的主导地位扩展到2008年的13家公共交易所。富有进取心的市场参与者意识到，他们可以创造一种新的竞争优势。2通过观察某个交易所发生的、可能“左右市场走势”并改变其他交易所特定商品价格的交易，他们可以利用几微秒的时间优势，创造一个小小的机会窗口。例如，如果一家大型养老基金要求其经纪人购买10万股微软股票，高频交易公司可能会在交易在一家交易所进行时检测到这笔交易的信号，并采取行动在其他交易所获利，方法是买入微软股票，然后立即以略高的价格卖给养老基金。对于养老基金来说，这笔交易的成本可能比预期略高，但每天数百万笔交易的重复交易，其税负将变得相当可观。

When stock exchanges began to process trades digitally and to proliferate in the 1990s and 2000s, expanding from the dominance of NYSE and NASDAQ to thirteen public exchanges by 2008, enterprising market actors realized that they could carve out a new kind of competitive advantage.² By observing trades occurring on one exchange that would “move the market” and changing the price of a particular commodity elsewhere, they could use the temporal advantage of a few microseconds to create a small window of opportunity. For example, if a major pension fund asked its broker to buy, say, 100,000 shares of Microsoft, HFT firms might detect the signals of this trade as it moved through one exchange and act to profit from it on others by buying up Microsoft stock with the intention of immediately selling it at a slightly higher price to the pension fund. For the pension fund, the trade would be just a little more expensive than anticipated, but iterated over millions of transactions a day the taxes become substantial.

高频交易（HFT）提供了一个最纯粹的算法案例，它从根本上改变了现有的文化机器——一个古老的数学、社会实践、基于信仰的社群和套利机制的集合体，我们称之为“市场”。高速交易和算法的引入，实际上是全自动的商业引擎，其作用远不止将人类从交易大厅中移除：这些系统在与人类以及彼此之间公开竞争，并且正在逐渐改变更广泛的资本流动。高频交易套利者通过复杂的地理策略来建立优势，例如，他们将服务器和光纤线路比竞争对手更靠近交易所的中央服务器几英尺，或者租用通信线路，将金融网络中两点之间最佳的直接信号路径缩短几英里。虽然他们是交易员，但他们几乎从未亏损，因为他们的套利方式始终依赖于更优厚的信息，而且他们不像更传统的经纪人那样套利。他们在交易结束时从不持有股票，只持有现金^。3

HFT offers one of the purest examples of algorithms that are fundamentally altering an existing culture machine, that venerable assemblage of math, social practices, faith-based communities, and arbitrage that we call “the market.” The introduction of high-speed trading and algorithms that are effectively fully automated engines of commerce has done more than eliminate humans from the trading floor: these systems operate in open competition with humans and one another, and they are gradually transforming the broader movement of capital. HFT arbitrageurs build their advantage through complex geographical maneuvers, by locating their servers and fiber-optic lines a few feet closer to the exchanges’ central servers than their competitors, or leasing communication lines that shave a few miles off the best direct signal pathway between two points on the financial grid. While they are traders, they almost never lose money, since their form of arbitrage always depends on superior information and they do not arbitrage risk like more traditional brokers. They never hold stocks at the end of the day, only cash.³

这是一种文化套利，而不仅仅是计算，因为高频交易利用了我们对买卖证券根本目的的假设。市价订单类型的激增（远超100种，远远超出了我们熟悉的“买入”、“卖出”和“限价”类型），买卖双方愿意使用特定交易协议的激励机制，甚至运营中的交易所数量的不断增长，所有这些都促成了交易量和交易速度的提升。4^这些系统不仅交易证券，还交易信息，这一直是市场的一个重要功能。但它们将这些信息原子化并加密，放入一系列新的黑匣子中——交换系统、交易匹配引擎以及算法本身——这使得高频交易公司以及与其打交道的华尔街主要公司能够创造一种底层的撇脂经济。这些为其持有者创造数十亿美元利润的算法，利用对网络延迟（即信号从一个节点发出到到达另一个节点之间的滞后时间）的特殊了解，在边缘地带运作。找到比下一个人领先一点点的方法，哪怕只是千分之一秒，也能让公司创建严密保护的金融捷径和伏击点，以便在他们通过金融网络时可以利用更传统的商业模式。

This is a form of cultural arbitrage, not just computation, because HFT takes advantage of our assumptions about the fundamental purpose of buying and selling securities. The proliferation of market order types (well over 100, ranging far beyond the familiar “buy,” “sell,” and “limit” types), the incentives for buyers and sellers willing to use particular trading protocols, and even the growing number of exchanges in operation have all contributed to an increased volume and pace of trading.⁴ Much more than securities, these systems trade information, which has always been an essential role of the markets. But they atomize and encrypt that information in a new series of black boxes—the switching systems, trade-matching engines, and algorithms themselves—that allow HFT firms and the major Wall Street firms that deal with them to create a kind of subaltern skim economy. The algorithms that generate billions in profits for their keepers operate at the margins by capitalizing on special knowledge of network latency, or the lag-time between when a signal originates at one node and arrives at another. Finding ways to get a little bit ahead of the next guy, even at the level of thousandths of a second, allows companies to create closely guarded financial shortcuts and ambush-points where more traditional modes of commerce can be exploited as they move through financial networks.

从其自身角度来看，高频交易使我们能够想象一个熟悉的文化机器——市场——如何将其转化为真正的计算空间。金融市场本身的规模和随机性是这些算法的主要画布，其决策依据关于定价、内在价值和信息速度的严格限制的抽象概念进行编码。这些系统运行的速度和规模，以及它们日益受到高盛等投资巨头的青睐，正在开始改变市场的根本行为^。5

Considered on its own terms, HFT allows us a way to imagine a familiar culture machine—the market—in its translation to a truly computational space. The volume and stochasticity of the financial markets themselves are the primary canvas for these algorithms, with decisions coded along tightly constrained abstractions about pricing, inherent value, and the pace of information. The speed and volume at which these systems operate, and their increasing cooption by investment titans like Goldman Sachs, are beginning to change the fundamental behavior of the markets.⁵

刘易斯故事中的英雄是那些试图消除高频交易算法“不公平”掠夺，并为证券交易创造一个公平竞争环境的人，他们设想证券交易就应该如此。他们的解决方案不是消除算法，而是消除那些可能找到新方法钻系统空子的人类中介。“不再需要任何人为干预……目标必须是消除任何不必要的中介。” ⁶对于华尔街的参与者来说，高频交易的问题不在于它在计算空间中的地位，而在于它与物质世界之间令人不安的联系。

The heroes of Lewis’s story are those trying to eliminate the “unfair” predation of HFT algorithms and create an equal playing field for the trading of securities as they imagine such things ought to be traded. Their solution is not the elimination of algorithms but instead of the human intermediaries who might find new ways to game the system. “There was no longer any need for any human intervention. … The goal had to be to eliminate any unnecessary intermediation.”⁶ For the players on Wall Street, the problem with HFT was not its position in computational space, but its troubling ties back to the material world.

刘易斯描述了他的改革者们创建的两张图表，它们旨在展示市场不断变化的本质：第一张图表展示了人类感知到的活动，其中一系列密集的交易事件在十分钟内逐秒记录。第二张图表则展示了一秒钟的算法视图，以毫秒为单位进行标记：

Lewis describes two charts his reformers created to demonstrate the changing nature of the market: the first shows activity as humans perceive it, with a crowded series of trading events marked second-by-second over the space of ten minutes. The second chart demonstrates an algorithmic view of a single second, marked off by milliseconds:

一秒钟内的所有市场活动都如此集中——仅仅1.78毫秒——以至于在图表上，它就像一座从沙漠中拔地而起的方尖碑。在98.22%的毫秒内，美国股市什么也没发生。对计算机来说，即使是世界上交易最活跃的股票市场，也只是一个平静无波、几乎昏昏欲睡的地方。……

“这个尖峰代表什么？”一位投资者指着方尖碑问道。

“这是你的订单之一，已经到货了，”布拉德说^。7

All the market activity within a single second was so concentrated—within a mere 1.78 milliseconds—that on the graph it resembled an obelisk rising from the desert. In 98.22 percent of all milliseconds, nothing at all happened in the U.S. stock market. To a computer, the market in even the world’s most actively traded stock was an uneventful, almost sleepy place. …

“What’s that spike represent?” asked one of the investors, pointing to the obelisk.

“That’s one of your orders touching down,” said Brad.⁷

突然飙升意味着一家大型机构投资者下单，这导致等待的金融算法陷入疯狂的抢先交易、反向竞价和仓位操作，所有这些都以人类无法处理的速度进行。然而，对于这些算法来说，它却慢得令人窒息，它们花了 98% 的时间等待某人采取行动：1.78 毫秒作为一个时间跨度几乎难以理解。相比之下，人类的平均反应时间大约为 200 毫秒。当代算法的速度已经超越了我们无需计算帮助就能应对的速度。

The sudden spike represented a large institutional investor placing an order, which then caused the waiting financial algorithms to leap into a frenzy of front-running, counter-bidding, and positioning, all conducted at a pace that is impossibly fast for human processing. Yet it was languidly slow for these algorithms, which spent 98 percent of that second waiting for someone to make a move: 1.78 milliseconds is more or less incomprehensible as a temporal span. By contrast, the typical human reaction time is something on the order of 200 milliseconds. The algorithmic contemporary has outpaced our capacity to manage it without computational help.

这种视觉叙事巧妙地捕捉了计算的空间，凸显了其截然不同的时间尺度这一本质特征，同时也表明我们很难想象这种时间性所带来的后果。活动的激增代表着订单，但这些订单背后是金融信号陷阱和传感器、用于确定行动方案的算法决策树、用于执行这些决策的交易工具，以及在难以想象的遥远未来，也许是几分钟或几小时后（如果真的发生的话），由人类来审查这些活动。沙漠中的方尖碑让人想起，如果西奥多有投资账户，《赫尔》中的萨曼莎会悲伤地称之为“交易之间的空间”。但它也展示了算法如何占领人类活动的边缘。斯科特·麦克劳德巧妙地用漫画术语“边缘”捕捉了叙事的叙事空间。8^边缘是指画面之间的空白，读者在将静态图像拼接成引人入胜、浑然一体的动画叙事时，通常会忽略它。正是在意识中，我们推断出那些在叙事表层话语中常常被间接提及的事件和行为。你或许会认为，意识是从人类认知难以触及的底层构建叙事的行为——我们如何将感官输入的碎片和推理串联成一段个人历史。

This visual narrative neatly captures the space of computation, highlighting the essential feature of its radically different time scale but also signaling how difficult it is for us to imagine the consequences of that temporality. The spike of activity represents orders, but behind those orders are financial signal traps and sensors, algorithmic decision trees to determine a course of action, trading tools to implement them, and, somewhere in the unimaginably distant future, perhaps minutes or hours later (if it happens at all), a human to review the activity. The obelisk in the desert is a reminder of what Her’s Samantha would mournfully call “the space between the trades,” if Theodore had an investment account. But it also demonstrates how algorithms are colonizing the margins of human activity. Scott McCloud wonderfully captures the extradiegetic space of a narrative with the comic book term “the gutter.”⁸ The gutter is the blank space between panels, largely ignored by the reader as she stitches static images into a compelling, animated narrative that feels seamless. It is where we infer events and actions that often are only obliquely referenced in the surface discourse of a narrative. You might argue that consciousness is the act of constructing a narrative from the inaccessible gutter of human cognition—the ways that we stitch together a personal history from fragments of sensory input and inference.

我们通过创建过去的模型模拟来实现这一点，从这些经验片段中插入时间的连续性。我们假装这些“边缘”并不存在，用感觉完整的个人叙述和记忆来填补它们。套利是另一种插值形式，我们模拟的是向前而不是向后的时间，从而有效地预测未来并加以利用。但试图理解计算空间则是一个完全不同的挑战，它要求我们不是向前航行，而是更深入地探索时间。麦克劳德的“边缘”概念揭示了实施的基本规则，甚至是物理学，如何具有认识论意义。当前全球金融的转型很大程度上取决于两个几乎完全互斥的时间宇宙之间金钱和信息的套利。刘易斯描述的方尖碑是算法交易的人性化隐喻，几乎就像一张来自计算宇宙的程式化明信片，在宇宙中时间就是一切。高频交易算法将“边缘”——例如下单和执行订单之间的差距——转化为竞争性计算的舞台。在叙事和财务的例子中，排水沟是时间转化为意义和价值的地方。

We do this by creating a model simulation of the past, interpolating temporal continuity from those experiential fragments. We pretend the gutters aren’t there, paving them over with personal narratives and memories that feel complete. Arbitrage is another form of interpolation where we simulate forward in time instead of backward, effectively predicting the future and capitalizing on it. But trying to understand computational space is a different challenge entirely, requiring us to voyage not forward but deeper into time. McCloud’s notion of the gutter reveals how the ground rules, even the physics, of implementation have epistemological implications. The current transformation of global finance largely depends on the arbitrage of money and information between two temporal universes that are almost completely mutually exclusive. The obelisk that Lewis describes is a human metaphor for algorithmic trading, almost like a stylized postcard from a computational universe where time is everything. HFT algorithms translate the gutter—the gaps between placing an order and executing it, for example—into an arena for competitive computation. In both the narrative and financial examples, the gutter is the place where time is translated into meaning and value.

说高频交易算法对不同地理位置的股票价格进行简单的比较套利，就是说它们重视时间，但更重要的是，它们重视时间过程。在这些交易中，相关证券的内在价值、其领导地位、市盈率都毫无意义。这种做法的意义来自于 A 点和 B 点之间的时间差距，以及利用他人交易获利的概率。它们以微秒为单位对特定时间和空间意识形态进行建模的效率正在逐渐重塑它们所处的整个生态系统。通过这种方式，高频交易算法用一种基于过程的新价值结构取代了证券交易中原有的价值结构，即将持有和交易的股票视为对未来经济成功的投资。

By saying that HFT algorithms perform a simple comparative arbitrage between stock prices at different geographic locations is to say that they value time, but, more important, that they value the temporal process. In these transactions, the intrinsic worth of the securities in question, their leadership, their price-to-earnings ratios, are all meaningless. The meaning of the exercise is derived from the temporal gap between point A and point B, and the odds of executing a trade that capitalizes on someone else’s. Their efficiency at modeling a particular ideology of time and space, denominated in microseconds, is gradually reshaping the entire ecosystem they inhabit. In this way HFT algorithms replace the original structure of value in securities trading, the notion of a share owned and traded as an investment in future economic success, with a new structure of value that is based on process.

重视文化

Valuing Culture

流程套利是谷歌商业模式的核心；这家全球最大的公司之一（现为 Alphabet Corporation）正是建立在对文化信息的评估之上。1996 年，拉里·佩奇和谢尔盖·布林在斯坦福大学创建了首个 PageRank 算法，标志着在线搜索这一基本问题的新纪元。布林和佩奇的创新之处在于利用网络本身固有的本体结构来评估知识。大学网站的页面可能比商业网站的页面更能提供信息；一篇已经被数百篇其他文章链接的新闻文章，比一篇在其他地方只有少量引用的新闻文章更有信息量。谷歌将快速扩张的互联网视为地铁系统（有些车站比其他车站更繁忙、更便捷），而非大海捞针，从而建立了一个数字宇宙的算法模型，该模型如今几乎塑造了数字文化的方方面面。

The arbitrage of process is central to Google’s business model; one of the world’s largest companies (now in the form of Alphabet Corporation) is built on the valuation of cultural information. The very first PageRank algorithms created by Larry Page and Sergei Brin at Stanford in 1996 marked a new era in the fundamental problem of search online. Brin and Page’s innovation was to use the inherent ontological structure of the web itself to evaluate knowledge. Pages on university websites were likely to be better sources of information than those on commercial sites; a news article that had already been linked to by hundreds of others was a stronger source of information than one that had only a few references elsewhere. By viewing the rapidly expanding Internet less like a haystack and more like a subway system (where some stations are much busier and more convenient than others), Google established an algorithmic model of the digital universe that now shapes almost every aspect of digital culture.

最初浏览网络的原始尝试依赖于人工干预：一个“每日酷炫网站”或第一个像数字奇珍屋一样策划内容的博客，创造了一种独特的、基于美学的回应，以应对那些基本上未被绘制的网络。9^但我们与互联网的关系逐渐从类似公告牌或报纸的东西（我们可能在其中偶然浏览信息）演变为一种提供金融、工业和个人服务（从银行账户到在线约会）的基本公用事业。当我们打开水龙头时，我们期望流出稳定的水，我们希望水是（无毒的）水，而不是“每日酷炫液体”。同样，我们现在期望互联网成为一种提供可靠、甚至可替代信息的公用事业。

The first primitive efforts to navigate the web relied on human intervention: a “cool site of the day” or the first blogs that curated content like a digital Wunderkammer, creating an idiosyncratic, aesthetically grounded response to those largely unmapped networks.⁹ But our relationship with the Internet gradually evolved from something like a bulletin board or a newspaper, where we might browse for information serendipitously, to an essential utility that provides financial, industrial, and personal services ranging from bank accounts to online dating. When we turn on the faucet, we expect a reliable stream of something that is, we hope, recognizable as (nontoxic) water, not a “cool liquid of the day.” So too, we now expect the Internet to serve as a utility that provides dependable, and perhaps fungible, kinds of information.

谷歌自1998年成立以来开发的PageRank及其补充算法，最初只是为尚处于萌芽阶段的互联网提供复杂的检索工具。但自那时起，谷歌及其网络的迅猛扩张已将其假设转化为合理化的法则，正如狄德罗的相互关联主题框架塑造了无数的百科全书、索引和列表一样。在21世纪初的“搜索大战”中，谷歌巩固了其主导地位，一个转折点出现了，PageRank的观测系统成为了网络文化结构中的一股决定性力量。如今，谷歌掌控着大约三分之二的在线搜索，一个蓬勃发展的“搜索引擎优化”行业正在利用和操纵谷歌的算法来吸引流量和广告^。10

PageRank and the complementary algorithms Google has developed since its launch in 1998 started as sophisticated finding aids for that awkward, adolescent Internet. But the company and the web’s spectacular expansion since then has turned their assumptions into rationalizing laws, just as Diderot’s framework of interlinked topics has shaped untold numbers of encyclopedias, indexes, and lists. At some point during the “search wars” of the mid-2000s, when Google cemented its dominance, an inversion point occurred where the observational system of PageRank became a deterministic force in the cultural fabric of the web. Google now runs roughly two-thirds of searches online, and a vibrant industry of “search engine optimization” exists to leverage and game Google’s algorithms to lure traffic and advertising.¹⁰

PageRank 的核心是对人类判断行为进行记录，统计数百万人对网络不同角落的链接和相对关注度。PageRank 的专利将该算法描述为“一种为链接数据库中的节点分配重要性等级的方法，例如任何包含引文的文档数据库、万维网或任何其他超媒体数据库。” ¹¹简而言之，这是一种在世界上创建等级制度的工具，不仅根据信息的固有特性，还根据重要性或受欢迎程度的特定定义对其进行排序。可以毫不夸张地说，从更原始的搜索引擎到 PageRank 及其模仿者的算法加权的转变已经深刻地改变了知识在社会中的作用，就像狄德罗和达朗贝尔在争论百科全书应使用哪种排序方案时所做的那样。

At its heart, PageRank catalogs human acts of judgment, counting up the links and relative attention millions of people have paid to different corners of the web. The patent for PageRank describes the algorithm as “a method [that] assigns importance ranks to nodes in a linked database, such as any database of documents containing citations, the World Wide Web or any other hypermedia database.”¹¹ In short, this is a tool for creating hierarchy in the world, sorting information not merely by its inherent qualities but by a certain definition of importance or popularity. It is not too much to argue that the transition from more primitive search engines to the algorithmic weighting of PageRank and its imitators has shifted the role of knowledge in society as profoundly as Diderot and d’Alembert did when they debated which ordering schema to use for the Encyclopédie.

这种转变被笼罩在这份名单的谦逊之中，这份名单是符号学家、哲学家和文学评论家翁贝托·埃科曾经称之为“文化的起源”的简单的知识结构：

That transformation is shrouded in the modesty of the list, that simple intellectual construct that the semiotician, philosopher, and literary critic Umberto Eco once identified as “the origin of culture”:

文化想要什么？让无限变得可理解。它也想要创造秩序——并非总是如此，但常常如此。作为人类，我们该如何面对无限？我们该如何尝试理解那不可理解的事物？通过清单、目录、博物馆的藏品、百科全书和词典^。12

What does culture want? To make infinity comprehensible. It also wants to create order—not always, but often. And how, as a human being, does one face infinity? How does one attempt to grasp the incomprehensible? Through lists, through catalogs, through collections in museums and through encyclopedias and dictionaries.¹²

百科全书的编纂者和谷歌都声称，他们的项目并非创造等级制度，而是对其进行建模——他们的知识本体论仅仅是对文化中既有结构的更有效的映射。然而，正如埃科所言，在这两种情况下，他们所创建的结构很快就成为了自身的排序机制，塑造了它原本旨在观察的文化空间。百科全书最重要的功能之一就是为了节省时间而压缩知识，将数百万人多年的洞见提炼成参考文本，从而加快研究和发现的步伐。对狄德罗和达朗贝尔来说，正是这本书在政治、哲学乃至认识论上为法国大革命铺平了道路。对谷歌来说，这是一个机器能够理解、交流和预测的未来——一个完整的知识计算本体论的未来。

Both the encyclopédistes and Google would argue that their projects do not create hierarchy but model it—that their knowledge ontologies are simply more effective maps for structures that already existed in culture. And yet, as Eco suggests, in both instances the structure they created quickly became an ordering mechanism of its own, shaping the cultural space it was designed to observe. One of the most vital functions of the encyclopedia was to compress knowledge for the sake of time, distilling millions of person-years of insight into a reference text that would accelerate the pace of research and discovery. For Diderot and d’Alembert, it was the book that politically, philosophically, and perhaps epistemologically paved the way for the French Revolution. For Google, it is a future of machines that understand, converse, and anticipate—a future of a complete computational ontology of knowledge.

这两个项目原本的雄心壮志如今已基本实现。但它们也是在历史时刻实施的系统，实现了不同的目标：重新构建文化价值和时间性。正如高频交易 (HFT) 算法利用了交易技术和文化构建中的差距一样，PageRank 套利了丰富的潜在信息领域——构成全球互联网的连接网络——并将其提炼为可立即访问的资源。谷歌搜索通过追踪与我们的查询相关的最有用资源（这是一种一级预期），从长远来看节省了我们的时间。但它也通过以惊人的即时性提供搜索结果，在短期内节省了我们的时间：谷歌很早就意识到，即使是十分之一秒的延迟也会降低用户参与度，并据此构建了其基础设施。13^在知识综合和即时满足层面上的时间套利是推广百科全书和搜索引擎的关键因素。

These were the two projects’ ambitions, now largely fulfilled. But they were also systems implemented in historical moments where they achieved something different: a reframing of cultural value and temporality. Just as the HFT algorithms exploit gaps in the technical and cultural construction of trading, PageRank arbitrages a rich field of latent information—the network of connections making up the global Internet—and distills it into an immediately accessible resource. Google Search saves us time in the long run by tracking down the most useful resources relevant to our query, a first-order form of anticipation. But it also saves us time in the short run by providing its search results with startling immediacy: Google realized early on that delays of even a tenth of a second would reduce user engagement, and built its infrastructure accordingly.¹³ The arbitrage of time at the level of knowledge synthesis and immediate gratification are crucial factors in promoting both encyclopedias and search engines.

这种时间套利不可避免地也导致了金融价值的重新协商：时间就是金钱，尤其是在数百万台服务器上以毫秒为单位的时间乘以1。为此，谷歌需要AdSense。从商业角度来看，PageRank为思想的传播创建了一个基本指数，是注意力经济中必不可少的货币。14当谷^{歌开始利用市场竞价系统AdSense在其搜索结果中出售广告时，它成功地将注意力以以前难以想象的规模货币化。 2013}年，谷歌的收入超过550亿美元，其中90%以上来自广告。15现在，该公司将帮助您注册域名、建立网站、分析流量并为网站投放广告，然后其算法将对网站进行索引和排名。2014年，谷歌的市值超过了埃克森美孚，成为全球市值第二高的公司，仅次于苹果^。16

That temporal arbitrage inevitably led to a renegotiation of value in financial terms as well: time is money, especially time measured by milliseconds multiplied across millions of servers. For this, Google needed AdSense. From a business perspective, PageRank creates a basic index for the circulation of ideas, an essential currency in the economy of attention.¹⁴ When Google began selling advertisements against its search results with the market bidding system AdSense, it succeeded in monetizing that attention at a previously unimaginable scale. In 2013, Google earned over $55 billion in revenue, of which more than 90 percent came from advertising.¹⁵ Now the company will help you register the domain name, build the website, analyze the traffic, and serve the ads for the site, which its algorithms will then index and rank. In 2014, Google exceeded the market capitalization of ExxonMobil, leaving it second only to Apple among the most valuable companies in the world.¹⁶

谷歌的典型广告收入仅为公司向客户提供服务的极小一部分，但在其每天投放的数百亿条广告中，这些零碎加起来构成了一种使用互联网的最低交易成本，由其最强大的“守门人”收取。17 ^AdSense的功能实际上本身就是一种高频交易套利：每当用户导航到通过谷歌网络投放广告的网站时，都会进行快速拍卖，竞标出价最高的营销人员获得其广告投放权。这些交易将我们每个人在网上留下的大量用户数据——根据我们的购买历史、人口统计数据和许多其他因素构建的详细消费者资料——商品化，以便广告商能够根据用户的“兴趣”以及托管网站的上下文空间来识别目标市场。但 AdSense 也是一种时间套利，它将时间商品化，就像高频交易系统利用几毫秒的延迟榨取利润一样有效。谷歌利用 AdSense 的即时性、通过将当代的商品化，即潜在客户真正徘徊在门槛上的那一刻，建立了巨大的业务。

The typical Google advertisement nets the company some tiny fraction of a penny to serve up to a customer, but over the volume of the tens of billions of ads it serves each day, those fractions add up to a kind of minimal transaction cost for using the Internet, collected by its most powerful gatekeeper.¹⁷ The functionality of AdSense is in fact a kind of HFT arbitrage in its own right: every time a user navigates to a site serving advertisements via Google’s network, a rapid auction takes place for the marketers with the highest bids to serve their ads. These transactions commoditize the long trail of user data each of us leaves behind online—the detailed consumer profiles about us informed by our purchase history, demographics, and many other factors—so that advertisers can identify target markets based on user “interests” as well as the contextual space of the host website. But AdSense is also a form of temporal arbitrage, commoditizing time just as effectively as HFT systems milking profits out of a few milliseconds of lag. Google has built a tremendous business out of the immediacy of AdSense, out of commoditizing the contemporary, the moment right now when a potential customer is literally hovering at the threshold.

互联网广告，如同所有广告一样，是一种文化潜伏期或时间使用税，对流动的注意力市场施加了轻微的阻力。这些微型拍卖令人难以置信的算法基础设施、缓存内容的魔力以及高度协调的全球套利，揭示了一个微小却可衡量的转化点：文化价值与商业价值之间差距的那几分之一秒。这种停顿显而易见，它不仅是一种延迟，更是一种持续，一种新的外部化记忆形式，充满了消费者欲望的算法模型。18^我们被那些尚未购买的鞋子、汽车和假期所困扰，远比营销公司精心维护的数字自我的阴影更直接。PageRank 和 AdSense 实际上是同一枚硬币的两面，将获取普遍知识的理想与完美理解我们每个人及其欲望的理想结合在一起。它们是谷歌将知识探索转化为算法过程的核心使命的两种表达。

Internet ads, like all advertisements, are a form of cultural latency or temporal use tax, placing minor drag on the fluid market of attention. The incredible algorithmic infrastructure of these micro-auctions, the magic of cached content and highly coordinated global arbitrage, reveals a tiny but measurable point of translation: the fraction of a second that marks the gap between cultural and commercial value. That pause is noticeable, not just as a delay but as a kind of persistence, a new form of externalized memory filled with algorithmic models of consumer desire.¹⁸ We are haunted by the shoes, the cars, the vacations that we have not yet purchased much more directly than we are by the hidden shadows of our digital selves that marketing companies carefully maintain. PageRank and AdSense are really two sides of the same coin, uniting the ideal of access to universal knowledge with that of perfectly understanding each of us and our desires. They are two expressions of Google’s central mission to translate the quest for knowledge into algorithmic process.

PageRank的套利，以及AdSense的流程，不仅呈现了一系列离散的计算事件，更呈现出一种关于文化价值的普遍叙事。这个故事的动力源于我们持续的参与和关注，源于谷歌的广告收入，也源于其作为几乎所有数字体验中心平台日益增长的角色。为了从不同的角度重塑价值，这些相互关联的套利系统体现了文化理论家艾伦·刘（Alan Liu）所说的“信息精神”：一种人文主义风格的风格，或者说，计算主义风格的抽象层次，它在信息本身之上构建了一个批判性的上层建筑。19用^刘的话来说，我们可以认为，网络已经从浪漫主义阶段进入理性阶段，告别了早期手工编码网站的自制崇高，取而代之的是如今占据主导地位的、精心设计的品牌化、自动化页面服务的企业秩序^。20

The arbitrage of PageRank, the process of AdSense, present not just a series of discrete computational events but a pervasive narrative about cultural value. It’s a story that is fueled by our continued engagement and attention, by Google’s advertising revenues, and by its growing role as a central platform for nearly all forms of digital experience. To recast value in a different light, these interlocking arbitrage systems embody what cultural theorist Alan Liu has called an “ethos of information”: a style, in humanistic terms, or a level of abstraction in computational ones, that creates a critical superstructure over the information itself.¹⁹ In Liu’s language we might argue that the web has moved from a romantic to a rational phase, leaving behind the homebrew sublime of early hand-coded sites for the elaborate corporate order of branded, automated page serving that dominates today.²⁰

换句话说，尽管谷歌的业务建立在精准绘制互联网文化本体的基础上，但其角色已逐渐从索引器转变为架构师。其核心商业模式通过将算法与消费文化有效结合而获得发展，并且越来越依赖于运用算法来深刻地定义这种文化。谷歌在搜索关键套利领域的主导地位，催生了对新型素养的迫切需求，最明显的体现是“星际迷航”计算机和谷歌搜索本身。该公司作为互联网首席架构师的新兴角色，本质上仍然是一种文化套利，因此也是一个暂时的项目。谷歌希望预测、预先确定其众多用户可能想要利用的每一种资源。随着公司在塑造数字文化整体永久档案方面变得更加有效，这需要新的记忆、隐私，甚至遗忘词汇。

In other words, while Google built its business on accurately mapping the cultural ontologies of the Internet, its role has gradually shifted from indexer to architect. Its core business model gained traction by effectively marrying algorithms to consumer culture, and it increasingly depends on using algorithms to define that culture in profound ways. Google’s dominance over the critical arbitrage of search has created a pressing need for new forms of literacy, most publicly in the context of the Star Trek computer and Google search itself. The company’s emergent role as chief architect of the Internet is still, at heart, a form of cultural arbitrage and therefore a temporal project. Google wants to anticipate, to predetermine, every possible resource its many users may want to tap. This requires new vocabularies of memory, of privacy, and even of forgetting as the company becomes more effective at shaping the holistic, permanent archive of digital culture.

这些新的素养在欧洲最具争议，因为在那里，隐私被视为一项固有的人权，而非契约的产物。最近，欧洲法院的判决迫使谷歌删除某些被判定对个人有害的搜索结果，因为这些搜索结果会凸显用户过去的负面事件。21^谷歌的回应是，在每一次删除后都加上自己的注释，提醒人们某些搜索结果已被审查，以维护埃里克·施密特令人难忘地表达的完全透明的理念：“如果你有一些不想让任何人知道的事情，也许你一开始就不应该这样做。” ²²但套利当然建立在保护和精心管理某些信息的时间优势之上：广告、浏览行为、搜索查询。Google Now 请求访问我们的搜索历史、实际位置和其他数据，以便提供其服务，作为回报，它承诺不仅组织用户现在的时间，还组织用户不久的将来的时间。它会根据交通状况建议何时出发参加下次会议，创建一个私密的个人提醒系统，利用公共和私人数据进行套利。随着我们逐渐意识到文化价值与算法套利深度交织所带来的后果，匿名和不可追踪的商业理念变得越来越有吸引力。

These new literacies have been most controversial in Europe, where privacy is understood as an inherent human right rather than the child of contracts. Recent court decisions there have forced Google to remove certain search listings judged damaging to individuals because it foregrounds negative episodes from their past.²¹ Google’s response has been to mark each erasure with its own annotation, alerting people that some results have been censored, to maintain the ideology of total transparency memorably expressed by Eric Schmidt when he said, “If you have something that you don’t want anyone to know, maybe you shouldn’t be doing it in the first place.”²² But of course arbitrage is built on the temporal advantage of keeping certain information protected and carefully managed: advertisements, browsing behaviors, search queries. Google Now requests permission to access our search histories, physical location, and other data in order to provide its services, and in return it promises to organize not just the present but the near future temporalities of its users. It will suggest when to leave for the next meeting, factoring in traffic, creating an intimate, personal reminder system arbitraging public and private data. As we come to grips with the consequences of the deep interlacing of cultural value and algorithmic arbitrage, the ideals of anonymity and untraceable commerce have become more and more appealing.

加密货币

Cryptocurrency

谷歌作为在线套利和文化估值的核心工具，其作用日益扩大，将互联网时代一些最美好的梦想变成了现实，但却以一种出人意料地悄无声息的幕后方式。1997年，未来学家彼得·施瓦茨和彼得·莱顿在数字革命期刊《连线》上发表了题为《长期繁荣》的文章，这是对那个时代愿景最著名的表达之一。23施瓦茨^围绕五波技术变革​​做出了许多预测，但《连线》杂志所描绘的超越性未来的一个持续转折点，​​是一种新的资本主义计算套利，它将商业从官僚监管、实物货币和主权控制的历史遗留问题中解放出来。施瓦茨所谓的“电子现金”的广泛应用，早在1994年和1996年就曾被《连线》杂志报道过，该杂志倡导后物质时代、全球网络经济，这为其关于信息胜利的更广阔愿景奠定了基础。早期的数字现金创业者意识到信息的价值日益增长，并试图创建一种从货币到信息的算法抽象。正如谷歌和其他当代科技领域的巨头所发现的那样，这种抽象更容易通过永恒的套利魔力反过来实现：从信息开始，然后将其转化为货币。

Google’s expanding role as a kind of central utility for arbitrage and cultural valuation online has brought some of the dot-com era’s fondest dreams to life, but in an unexpectedly quiet, backroom way. The futurists Peter Schwartz and Peter Leyden offered one of the best-known expressions of that era’s visions in 1997 with “The Long Boom,” published in that bulletin of the digital revolution, Wired magazine.²³ Schwartz offered many predictions centered on five waves of technological change, but a persistent pivot point for the transcendental Wired future was a new kind of computational arbitrage for capitalism that freed commerce from the historical relics of bureaucratic regulation, physical specie, and sovereign control. The widespread adoption of what Schwartz called “electronic cash” was something Wired had covered earlier in 1994 and 1996, and the magazine championed the postmaterial, global network economy that underpinned its broader vision about the triumph of information. Early digital cash entrepreneurs recognized the growing value of information and attempted to create an algorithmic abstraction from money to information. As Google and the other major players in the contemporary technology space discovered, the abstraction is much easier to work the other way through the timeless magic of arbitrage: start with information and turn it into money.

然而，正如我们所见，这种计算魔法依赖于庞大的观察和数据收集系统：公司积累了海量数据，从个人用户资料到互联网本身的深度索引。然后，访问这些宝库就成了个人消费者和试图触达这些消费者的营销人员的商品。在每个阶段，套利都会让这些公司从每笔交易、每笔广告微拍卖和信用卡消费中分得一杯羹。这些转变很容易被忽视：iTunes 上待售专辑的列表与我们在唱片店可能遇到的选择并无太大区别；Netflix 上的节目似乎仍然是一系列关于观看内容的选择。算法后端的巨大变化被包装成相对熟悉的消费者术语。换句话说，《连线》杂志所设想的商业革命，实际上是一场改革——并非没有动荡和伤亡，但在很大程度上是一场没有对主要行业造成影响的转型。

As we’ve seen, however, that form of computational magic depends on large systems of observation and data collection: companies amass vast troves of data, from individual user profiles to deep indexes of the Internet itself. Then access to these troves becomes a commodity for both individual consumers and the marketers seeking to reach those consumers. At every stage, arbitrage nets these companies a share of every transaction, every advertising micro-auction and credit card purchase. These shifts are easy to overlook: the list of albums for sale on iTunes is not so different from the choices we might have encountered in a record store; the offerings on Netflix still appear to be a series of choices about what to watch. The dramatic changes on the algorithmic backend are packaged in relatively familiar consumer terms. In other words, what Wired imagined as a commercial revolution has instead been a reformation—not without its upheavals and casualties, but largely a transformation that has left major industries intact.

这种套利巨变的成功本身就加剧了一些人的反对，他们认为我们参与的数字交易并非为了方便——例如，为了观看少量定向广告而提供免费服务——而是为了新的集体算法设计的价值体系而彻底剥夺个人隐私和自主权。或许，这种意识形态反应最典型的例子就是名为比特币的新型加密货币的迅速普及。

The very success of this arbitrage sea change has accentuated the objections of those who see the digital transactions we all participate in not as matters of convenience—free services provided in exchange for viewing a few targeted ads, for example—but as the radical evisceration of individual privacy and autonomy for the sake of new collective, algorithmically engineered systems of value. Perhaps the single greatest example of this ideological reaction is the rapid popularization of a new cryptocurrency called Bitcoin.

比特币最初源于一篇由看似虚构的数学家中本聪（Satoshi Nakamoto）于2008年11月发表的论文（大约在施瓦茨预测电子现金将成为主流的十年后）。在论文中，中本聪提出了一种新的金融模式，该模式将消除传统金融系统的关键漏洞：“一种基于密码学证明而非信任的电子支付系统，允许任何两个有意愿的参与方直接进行交易，而无需可信的第三方。” ²⁴这篇直截了当的论文描述了一种纯粹基于计算能力和数学（我将在下文中更详细地描述）的货币兑换系统，该系统不依赖于中央银行、正式的发行机构或其他传统货币的“信仰和信用”标准。如果说本书迄今为止一直被黑匣子所主导，那么比特币则声称是一个“玻璃盒子”，一个功能完全透明的防篡改系统。与其他开源平台一样，它建立在这样的逻辑之上：最佳安全性源于允许任何人检查其代码并提出改进建议。

Bitcoin first emerged as a paper published in November 2008 by the apparently fictional mathematician Satoshi Nakamoto (about ten years after Schwartz predicted e-cash would become mainstream). In the paper Nakamoto argued for a new financial model that would eliminate the key vulnerability of traditional financial systems: “an electronic payment system based on cryptographic proof instead of trust, allowing any two willing parties to transact directly with each other without the need for a trusted third party.”²⁴ The straightforward paper describes a system for exchanging currency based purely on computing power and mathematics (which I describe in more detail below), with no dependence on a central bank, a formal issuing authority, or other “faith and credit” standards of traditional currencies. If this book has been dominated by black boxes so far, Bitcoin purports to be a “glass box,” a tamper-proof system whose functions are entirely transparent. Like other open source platforms, it is founded on the logic that the best security comes from allowing anyone to inspect its code and suggest improvements.

比特币作为一种自由意志主义，甚至无政府主义的回应，受到了广泛关注，它对更成熟的金融和政治力量对贸易和货币施加的限制做出了回应，并因成为各种网络犯罪活动的首选货币而声名狼藉。但中本聪论文中最引人注目的影响才刚刚开始显现：一种颠覆上述等式的新型算法套利模型。该模型利用了公共与私人、身份与社群、文化与商业套利之间同样活跃的交易，但它颠覆了个人能动性与价值建构之间的关系。

Bitcoin has received significant attention as a libertarian, even anarchist response to the restrictions on trade and currency imposed by more established financial and political actors, and it’s gained notoriety as the currency of choice for all manner of criminal activity online. But the most compelling effect of Nakamoto’s paper is just beginning to emerge: a new model of algorithmic arbitrage that inverts the equation described above. This model takes advantage of the same brisk trade between public and private, between identity and community, between culture and commercial arbitrage, but it reverses the relationship between individual agency and the construction of value.

为了论证这一点，我首先从算法的角度解读比特币，它既是一个计算平台，也是一种意识形态——一个基于密码学家假设的完全偏执和不信任的系统。由于任何安全系统都可能被攻破，依赖任何第三方或外部受信任实体共享信息不可避免地会给系统带来漏洞。依赖银行转账需要与全球金融网络、错综复杂的监管数据收集要求、人员和电子媒体打交道，而所有这些都可能被操纵、入侵或监控。比特币通过两种关键的算法解决方案来应对这些挑战，一种是成熟的，另一种是极具创新性的。

To make this case, let me start with an algorithmic reading of Bitcoin as both a computational platform and an ideology—a system based on the cryptographer’s assumption of total paranoia and distrust. Since any security system can be compromised, relying on any kind of third party or externally trusted entity to share information inevitably introduces vulnerability into a system. Depending on a bank to transfer funds requires engaging with the global financial network, a complex thicket of regulatory data-gathering requirements, human beings, and electronic media, any of which might be manipulated, compromised, or surveilled. Bitcoin responds to these challenges with two crucial algorithmic solutions, one well-established and the second radically innovative.

比特币首先依赖于非对称加密算法，这种算法已成为所有数字安全应用的首选。这些算法依赖于“单向函数”——易于执行但极难逆向的计算。一个众所周知的例子（广泛使用的 RSA 加密算法的基础）：将两个大素数相乘得到一个巨大数很容易，但将一个巨大数分解成两个组成素数则要困难得多。通过依赖这些实际上不可逆的计算过程，比特币为其用户提供了一种验证其交易合法性的方法，而无需了解交易双方的身份。在素数的例子中，“私钥”由两个素数组成，用于对用户希望保护的数据进行数字签名。然后，可以使用从巨大数（两个素数的乘积，或者更一般地说，单向计算函数的输出，无论它是什么）派生出的“公钥”，通过确保私钥确实用于签名来验证数据。

Bitcoin depends first of all on asymmetric encryption algorithms, which have become the overwhelming favorite for all digital security applications. These algorithms rely on “one-way functions”—calculations that are easy to perform but extremely difficult to reverse-engineer. A well-known example (the basis of the widely used RSA encryption algorithm): it’s easy to multiply two large prime numbers together into a giant number, but much harder to factor a giant number into two constituent primes. By relying on these effectively irreversible computational processes, Bitcoin provides its users a way to authenticate that its transactions are legitimate without knowing anything about the parties to those transactions themselves. In the prime number example, a “private key” would be made up of the two prime numbers and used to digitally sign the data that a user wishes to keep secure. Then a “public key” derived from the giant number (the product of the two primes, or more generally the output of the one-way computational function, whatever it is) could be used to verify that data by ensuring that the private key was indeed used to sign it.

到目前为止，比特币只是另一种支付方案，依赖于某个中央机构来追踪公钥，并防范所谓的“双重支付问题”——即你刚刚收到的款项可能已经在其他地方被使用，类似于某种数字伪造。但比特币的第二个创新之处在于，我们在区块链的共识驱动机制中发现了一种新的计算套利形式。

Up to this point, Bitcoin would simply be another payment scheme that depended on some central authority to track public keys and defend against what is called the “double spending problem”—the risk that the money you have just received in payment might also have been spent somewhere else, analogous to a sort of digital counterfeiting. But Bitcoin’s second innovation is where we discover a new form of computational arbitrage, in the consensus-driven mechanism of the blockchain.

区块链是比特币历史上所有交易的公共账本。它详细记录了自比特币诞生以来的每一笔交易，这是一个超过 20GB 的数字文件，每个比特币软件客户端都必须在本地下载。区块链是一种算法，它实现了对比特币的政治批判，是一种套利的奇迹，颠覆了传统的隐私与透明度之间的关系。每一笔交易，每一个增量单位的比特币价值，都通过区块链进行追踪，并且每笔交易都与一个或多个买家和卖家的身份相关联。这些身份仅仅是从非对称加密协议（椭圆曲线数字签名算法）的公钥派生出来的字母数字字符串。这些交易记录了一定数量的比特币从一个身份转移到另一个身份的过程。通过区块链（该区块链由比特币社区不断更新和验证，我将在下文讨论），可以追溯每个货币单位的起源点，追溯它参与的每一笔交易。整个比特币市场都是公开的，从恐怖主义融资的最黑暗角落到预订酒店房间，从 Zynga 购买虚拟商品，以及从臭名昭著的数字市场丝绸之路订购大麻^。25

The blockchain is the public ledger of all Bitcoin transactions in the history of the currency. It contains a detailed accounting of every transaction since the currency’s instantiation, a digital file that now exceeds 20 gigabytes in size and must be downloaded locally by every Bitcoin software client. The blockchain is the algorithm that implements the political critique of Bitcoin, a marvel of arbitrage that inverts the traditional relationship between privacy and transparency. Every transaction, every incremental unit of Bitcoin value, is traced through the blockchain, and each of those transactions is tied to one or more buyer and seller identities. The identities are simply alphanumeric strings derived from the public keys of an asymmetric encryption protocol, the Elliptic Curve Digital Signature Algorithm. The transactions annotate the movement of some quantity of Bitcoin from one identity to another. Through the blockchain, which is constantly updated and authenticated by the Bitcoin community (as I’ll discuss below), it’s possible to trace each unit of currency back to an origination point, through every single transaction it’s ever been part of. The entire Bitcoin marketplace is an open book, from the darkest recesses of terrorism financing to booking hotel rooms, purchasing virtual goods from Zynga, and ordering marijuana from the infamous digital marketplace Silk Road.²⁵

但这究竟是如何运作的呢？由于比特币网络没有中央权威机构，任何完成交易的人都会通过点对点网络进行公告。该系统的去中心化特性旨在解决信息在网络中流动不均衡的问题，例如某些节点可能突然出现或消失，以及有意取消中央银行或交换站来关联和排序所有金融活动的设计约束。

But how does this actually work? Since the Bitcoin network has no central authority, anyone completing a transaction announces it through a peer-to-peer network. The decentralized nature of the system is meant to account for the problem that information may flow unevenly across the network, that some nodes may suddenly appear or disappear, and for the intentional design constraint of abolishing the central bank or switching station to correlate and sequence all financial activity.

这些不同的交易公告被比特币“矿工”打包成交易区块，然后他们竞争组装这些交易，并根据比特币现有的公共历史记录进行验证。这种劳动的成果是区块链的新区块。为此，他们必须解决一个任意且高度复杂的数学问题。第一个正确解答该问题的矿工将“赢得”该区块。此外，还有奖励：每个新区块中的第一笔交易是“生成交易”，它会创建一定数量的新比特币（数量会随着时间的推移逐渐减少）。解决该区块的矿工因投入了最多的计算资源来组装区块链的最新区块而获得此奖励（图 5.1）。矿工还通过索取处理这些不同交易的少量交易费来接受二次奖励（此费用随着时间的推移逐渐增加）。然后，网络中的其他节点接受这个新铸造的区块链尾部，并转向将新交易组装成一个新的区块。比特币软件经过精心校准，以便社区大约每十分钟生成一个新区块（就像中本聪的论文所建议的那样），比特币的整体生产本身也是精心策划的。一旦创建了 2100 万个比特币，系统将逐渐将完成新区块的奖励减少到零，从而完全停止创建新的比特币。26^到那时，仅交易费就足以激励比特币用户将他们的计算机用于无休止地更新区块链。

These different transaction announcements are bundled up into transaction blocks by Bitcoin “miners,” who then compete to assemble and validate these transactions against the extant communal history of the currency. The outcome of that labor is a new block for the blockchain. To do this, they must solve an arbitrary and highly complex math problem. The miner who is the first to correctly solve the problem “wins” that block. And there is a reward: the first transaction in each new block is a “generation transaction,” that creates a quantity of new Bitcoins (the number gradually decreases over time). The miner who solves the block earns this reward for throwing the most computational resources at assembling the latest block for the blockchain (figure 5.1). The miner also accepts a secondary reward by claiming a small transaction fee for processing these various trades (this fee gradually increases over time). Other nodes in the network then accept this newly minted tail for the blockchain and turn to assembling new transactions into a new block. The Bitcoin software is carefully calibrated so that the community generates a new block approximately every ten minutes (just like Nakamoto’s paper suggests), and the overall production of Bitcoin is itself carefully orchestrated. The system will gradually taper the reward for completing new blocks to zero, thereby ceasing the creation of new Bitcoins entirely, once 21 million Bitcoins have been created.²⁶ At that point, transaction fees alone will provide the incentive for Bitcoin users to dedicate their computers to endlessly updating the blockchain.

我之所以如此详细地阐述这一复杂的金融共识达成过程，是因为比特币不仅仅是一种去中心化的货币，更是以算法的方式对商业的重新评估。比特币真正的激进之处在于，区块链将其权威性建立在集体计算之上，并将其作为一种内在的价值形式。要理解这种转变，我们需要将比特币置于资本主义历史价值主张的背景下思考。正如卡尔·马克思的著名论述，工业资本主义建立在一种强大的抽象模式之上，这种模式将个人与其劳动利润分离，创造出一种异化形式，将个人的劳动抽象为可替代的商品和服务。一切都变得商品化，交换价值的抽象化逐渐掩盖了所有其他价值衡量标准。从很多方面来看，交换价值就是资本主义：股票交易所和市场经济所体现的符号账本。如果这种账本被区块链取代，我们不仅仅是在这些市场中引入了一种新的货币：我们正在创造一种新的账本，一种新的商品、服务和活动估值方式。

I am dwelling on the details of this elaborate process for delivering financial consensus because Bitcoin is not simply a decentralized currency but a revaluation of commerce in algorithmic terms. Bitcoin’s true radicalism stems from the fact that the blockchain grounds its authority on collective computation as an intrinsic form of value. To understand this shift we need to consider Bitcoin in the context of the historical value propositions of capitalism. As Karl Marx famously argued, industrial capitalism is based on a powerful mode of abstraction, one that separates individuals from the profits of their labor, creating a form of alienation that abstracts the work of individuals into fungible goods and services. Everything becomes commoditized, and the abstraction of exchange-value comes to obscure all other measures of worth. Exchange-value in many ways is capitalism: the symbolic ledger embodied by the stock exchange and the market economy. If that ledger gets usurped by the blockchain, we are not simply introducing a new currency into those markets: we are creating a new ledger, a new way of valuing goods, services, and activities.

许多人认为比特币的颠覆性力量仅仅源于去中心化。事实上，它确实用一种新的法定货币取代了国家的法定货币，这种货币没有金本位、没有保证的赎回政策或内在效用（燃烧比特币并不能保暖，而挖矿往往会使所涉及的处理器产生大量热量）。27^这绝非新鲜事：几千年来，商业文化在没有中央集权的情况下一直使用相互接受的价值衡量标准（箭头、贝壳），通常使用本身实际上毫无价值的标记或物品——并在必要时用一种系统取代另一种系统。这些系统依赖于对贸易社区的隐性信任：如果我今天接受这笔钱作为支付，明天我就可以在其他地方花掉它。金钱是我们集体注入信仰的符号，是一种依赖于抽象层面通过文化传递价值的技术。硬币、纸片或宝贝壳作为实物可能没有太多的内在价值，但它们是象征体系的标志，金融作家兼活动家布雷特·斯科特 (Brett Scott) 将其描述为类似于资本主义的机器代码——一种与债券、精品高频证券交易和信用违约掉期的果仁蜜饼层相去甚远的基础抽象，以至于毋庸置疑和看不见。28^金钱之所以有效，是因为我们都同时相信它，尽管历史上充斥着恶性通货膨胀、金融崩溃和严重腐败的事件，这种信仰可能在一夜之间消失。

Many people perceive Bitcoin’s disruptive force to derive merely from decentralization. It does, indeed, replace the fiat currency of the state with a new fiat currency, one backed by no gold standard, no guaranteed redemption policy or intrinsic utility (burning Bitcoins won’t keep you warm, though mining them does tend to generate a lot of heat from the processors involved).²⁷ This is by no means novel: mercantile cultures have used mutually accepted measures of value (arrowheads, seashells) for millennia without centralized authorities, often using markers or objects that are effectively valueless on their own—and replacing one such system for another when necessary. These systems depended on implicit trust in a trading community: if I accept this money in payment today, I will be able to spend it elsewhere tomorrow. Money is a symbol that we collectively infuse with belief, a technology for transmitting value through culture that depends on a layer of abstraction. The coins, pieces of paper, or cowrie shells may not have much intrinsic value as physical objects, but they are markers of a symbolic system that financial writer and activist Brett Scott described as analogous to the machine code of capitalism—a foundational abstraction so far removed from the baklava layers of bonds, boutique high-frequency securities trades, and credit default swaps as to be unquestioned and invisible.²⁸ Money works because we all believe in it at the same time, even though history is rife with incidents of hyperinflation, financial collapse, and gross corruption where that faith can vanish overnight.

国家货币和中央银行在等式中增加了第二层信任和定向管理，用国家权威来支持硬币或纸币的价值，在美国，这有时被称为联邦政府的“完全信任和信用”。29^货币可以得到内在支持，就像艾萨克·牛顿称量和追踪每枚英国硬币中的贵金属一样，也可以得到外在支持，就像政府维持金条储备以锚定其纸币的价值一样。一个国家及其中央银行可以控制新货币的流通，操纵汇率，强制固定商品价格，或者简单地通过法令改变其价值（就像朝鲜在 2009 年一夜之间发行新货币一样）。在这些情况下，货币不再仅仅依赖于支持流通的信仰共同体，还依赖于基于国家权力和宣称的智慧的第二个交叉共同体。

State currencies and central banks added a second layer of trust and directed management to the equation, backing the value of a coin or note with the authority of the state, what in the United States is sometimes called the “full faith and credit” of the federal government.²⁹ Currencies could be supported intrinsically, like Isaac Newton weighing and tracking the precious metals in each British coin, or extrinsically, like a government maintaining a reserve of gold bullion to anchor the value of its paper currency. A state and its central bank might control the introduction of new specie into circulation, manipulate exchange rates, mandate fixed prices for commodities, or simply change its value by fiat (as North Korea did in 2009 by introducing a new currency overnight). In these instances, currency no longer depends solely on the community of belief supporting circulation, but also on a second intersecting community based on the power and asserted wisdom of the state.

然而，信任正是比特币所反对的。从人性层面来看，用户仍然使用比特币进行交易，因为他们对社区充满信心（也就是说，他们相信自己能够将比特币出售给其他人），但该系统本身却建立在一个截然不同的原则之上。普通金融交易最终由国家行为体（例如，在美国，通过美联储和联邦存款保险公司等机构）支持，而比特币的终极基础原则是计算以及中本聪对交易的无信任愿景。我们得到的不是国库券或囤积的金条，而是区块链。区块链依赖于计算性的法币，奖励那些在计算每个新区块时贡献最大计算能力的矿工。该系统依赖于解决本质上毫无意义但却充当公共竞争角色的“工作量证明”问题，从而创造一种依赖于为比特币网络服务而消耗的计算机周期的价值形式。资本主义的核心抽象本身已被抽象为计算：这是对机器代码的升级，引入了新的价值基础层。

Yet trust is exactly what Bitcoin has positioned itself against. On a human level users still transact with Bitcoin because of their faith in a community (that is, faith that they will be able to sell their Bitcoins to someone else), but the system itself is founded on a very different principle. Where normal financial transactions are ultimately backed by state actors (in the United States, for example, through institutions like the Federal Reserve and the Federal Deposit Insurance Corporation), Bitcoin’s ultimate foundational principle is computation and Nakamoto’s trust-free vision for exchange. Instead of Treasury notes or stockpiled bullion, we get the blockchain. The blockchain relies on a computational fiat by rewarding the miners who bring the most computational power to bear on calculating each new block. The system depends on solving “proof-of-work” problems that are essentially meaningless but that serve as a public competition, creating a form of value dependent on the expenditure of computer cycles in the service of the Bitcoin network. The central abstraction of capitalism has itself been abstracted into computation: an upgrade to the machine code that introduces a new foundational layer of value.

这种阶段性变化最明显的表现是系统对某种算法民主或不可抗力的依赖，这取决于个人观点。区块链规则将“真相”地位授予第一个解决每个新的工作量证明问题的矿工，该解决方案随后被接受为整个系统的新公共记录。正如中本聪的论文所论证的那样，“只要大部分 CPU 能力由不合作攻击网络的节点控制，它们就会生成最长的链并超越攻击者。” ³⁰这修改了上述价值主张：价值不仅抽象为计算能力，而且抽象为集中的计算能力——要么用于比特币社区的利益以超越攻击者，要么被攻击者用来压倒平台。无论谁能首先解决毫无意义的数学问题，都将成为整个系统合法性的下一个认证者。

This phase change is most visible in the system’s dependence on a kind of algorithmic democracy or force majeure, depending on one’s perspective. The rules of the blockchain award “truth” status to the miner who is the first to solve each new proof-of-work problem, and that solution is then accepted as the new public record for the system as a whole. As Nakamoto’s paper argues, “As long as a majority of CPU power is controlled by nodes that are not cooperating to attack the network, they’ll generate the longest chain and outpace attackers.”³⁰ This modifies the value proposition above: value is abstracted not merely to computational power, but to concentrated computational power—either used for the benefit of the Bitcoin community to outpace attackers, or used by attackers to overpower the platform. Whoever can solve the meaningless math problem first serves as the next authenticator of the whole system’s legitimacy.

中本聪系统的稳定性依赖于持久的民主多数。作为一个经典的开源平台，比特币活跃的开发者社区不仅改进代码，还通过一种集体的粉丝参与来支持项目，通过专注的计算来相互强化项目的理念。每个新区块的加入都是整个系统的关键智力支点，它取决于争夺工作量证明方案的竞争，最终胜出者是“好人”，而这个系统不会试图以欺诈的方式篡改官方交易记录。为了实现这一点，大多数比特币矿工会集中他们的计算资源，将工作量证明问题分配到数千台不同的机器上，以提高计算速度，从而提高“赢得”区块的几率。事实上，即使是这种议会式的金融身份验证模式也存在风险：如果任何一个挖矿团体发展壮大，占据了解决区块链问题的整体计算池的50%以上，他们就会成为一个拥有随意篡改交易记录权力的多数派。比特币只有通过联合计算能力的平衡才能生存。

Nakamoto’s system depends for stability on a persistent democratic majority. As a classic open source platform, Bitcoin’s active community of developers improve the code but also support the project through a kind of collective fan engagement, a mutual reinforcement of the project’s ideology through dedicated computation. The addition of each new block to the chain is the crucial intellectual fulcrum for the entire system, and it depends on the race for a proof-of-work solution to be won by one of the “good guys,” a system that is not trying to fraudulently alter the official record of transactions. To make this work, most Bitcoin miners pool their computational resources, dividing up the proof-of-work problem across thousands of different machines to increase their speed of computation and thereby improve their chance of “winning” a block. Indeed, even this parliamentary model of financial authentication runs risks: if any of these mining collectives grows too large, taking over more than 50 percent of the overall computational pool tackling the blockchain problem, they will become a majority faction with the power to alter the transaction record at will. Bitcoin only survives through a federated balance of computational power.

比特币理论上无需信任的系统最终需要两种不同的信任：首先，对算法本身的信任，尤其是对其在区块链中透明基础的信任。其次，比特币鼓励参与者联合起来组成计算集体，围绕工作量证明解决方案的任意计算创建一个共享社区。在许多方面，这与刘易斯在《快闪小子》中描述的高频交易交易者的无需信任的世界非常相似。一个充满活力的算法生态系统依赖于一个善意的社区，这个社区会按照既定规则（无论这些规则多么残酷）进行竞争，并阻止对这些规则的任何更改。高频交易公司可能会找到越来越快地执行交易的方法，但减慢市场速度或改变交易的基本规则（例如，要求买家在出售股票前至少持有 24 小时）不符合任何人的利益。

Bitcoin’s notionally trust-free system ends up demanding two different kinds of trust: first, faith in the algorithm itself, especially in its transparent underpinnings in the blockchain. And second, Bitcoin encourages participants to band together into computing collectives, creating a shared community around the arbitrary calculation of proof-of-work solutions. In many ways, this is remarkably similar to the trust-free world of HFT traders that Lewis describes in Flash Boys. A vibrant algorithmic ecology depends on a community of goodwill to compete according to established rules (however cutthroat they may be) and discourage any changes to those rules. HFT companies may find ways to execute trades more and more quickly, but it is in nobody’s interest to slow markets down or change the ground rules of transactions (for instance, by requiring a buyer to hold a stock for at least twenty-four hours before selling it).

相似之处源于比特币，尽管它自诩为民主，但本质上是一个技术精英体系，将计算价值融入货币的资本主义基础。与传统货币类似，比特币的强大程度取决于其最强大的支持者（这里指的不是国家，而是最大的挖矿联盟）。但如今，定义其实力的核心价值是一种非常具体的资源：计算能力、硅片、电力。虽然游戏对任何人都开放，但真正能够影响结果的玩家是那些组成复杂挖矿联盟的人（这反过来又需要信任联盟公平地分享收益）。比特币的日常用户几乎不知道交易是如何计算的，也不知道这种权力平衡会如何影响他们自己的投资和交易。

The similarity arises because Bitcoin, despite its pretensions to democracy, is fundamentally a technologically elite system that inserts computational value into the capitalistic foundations of currency. Much like a traditional currency, Bitcoin is only as strong as its strongest proponent (in this case not a state, but the largest mining collectives). But the core value defining strength is now a very specific resource: computational power, silicon, electricity. While the game is open to anyone, the only players who can truly influence the outcome are those who band together in sophisticated mining collectives (which in turn requires trusting the collective to share the rewards fairly). The everyday users of Bitcoin have little idea how transactions are calculated or how this balance of power can affect their own investments and transactions.

最后，比特币这种程序化货币在将金融价值抽象为计算价值时，也产生了重要的外部性。正如爱德华·卡斯特罗诺瓦 (Edward Castronova) 所说，如果大量人开始使用没有传统主权国家支持的虚拟货币，可能会造成“普遍的动态衰退状态”，即那些仍在使用美元的人将不得不承担越来越多的税收义务来支持国家基础设施建设。31^从本质上讲，这就是美元的价值所在：美国这个全球超级大国，拥有强大的军事实力、道路、社会项目、监狱和养老金。用比特币取代这些，将有效地创建一个税收体系，支持一个不断扩张的计算周期网络，这些计算周期致力于挖掘区块链中的下一个区块。

Finally, the programmed currency that is Bitcoin also creates important externalities as it abstracts financial worth into computational value. As Edward Castronova argues, if significant numbers of people started using virtual currencies without the backing of traditional sovereign states, it could create “a general dynamic state of decline” where those still using dollars would have to take on ever-increasing tax obligations to support the infrastructure of the state.³¹ That, at heart, is what the dollar values: the global superpower of the United States of America with all its military might, its roads, its social programs, its prisons and pensions. Replacing that with Bitcoin would effectively create a tax system supporting an ever-expanding network of computational cycles dedicated to mining the next block in the chain.

可编程文化

Programmable Culture

比特币最近在主流市场取得了成功，从 Overstock 和亚马逊到各种酒店、旅游网站和小型企业，这表明它的计算价值逻辑正在流行，尤其是在其他算法驱动的企业中。正如斯坦福大学企业家巴拉吉·斯里尼瓦桑所说：“如果互联网是可编程通信，那么比特币就是可编程货币。” ³²与其说它是一种新的交易单位，不如说是一个新的金融计算平台。更深层次地说，它是一种意识形态，旨在使计算成为通用溶剂、通用价值指标，将《连线》杂志对商业数字化的想象延伸到它的逻辑结论。与谷歌的 PageRank 算法一样，比特币通过特定的、受时间约束的流程来定义价值，这些流程实现算法套利的形式。比特币的工作量证明在公共和私有之间创建了一种反转机制，恰恰逆转了 PageRank 的作用。对于比特币来说，其链接网络，即金融交易的复杂架构，是公开记录，但该网络的重要性却被掩盖：追溯每个比特币的历史记录、逐个区块识别它很容易，但要找出该交易单位的购买记录却极其困难。PageRank 则恰恰相反，它构建了一张精心守护的秘密网络地图，然后将其解读为对用户有意义的答案。这种镜像关系暗示着，随着我们将算法更深入地融入人类文化的核心机制，一系列更广泛的变化似乎正在浮现。

Bitcoin’s recent mainstream success, from Overstock and Amazon to a variety of hotels, travel sites, and smaller businesses, indicates that its logic of computational value is catching on, particularly with other algorithmically driven businesses. As Stanford entrepreneur Balaji Srinivasan argues: “If the Internet was programmable communication, Bitcoin is programmable money.”³² It is not so much a new unit of exchange as a new platform for financial computation. More profoundly, it is an ideology for making computation the universal solvent, the universal metric of value, extending the digitization of commerce Wired imagined to its logical conclusion. Like Google’s PageRank algorithm, Bitcoin defines value through particular, temporally constrained processes that implement forms of algorithmic arbitrage. Bitcoin’s proof of work creates an inversion mechanism between public and private that precisely reverses what PageRank does. For Bitcoin the network of links, the intricate architecture of financial transactions, is public record, while the significance of that network is obscured: it’s easy to trace the history of every Bitcoin, to identify it block by block, but extremely difficult to find out what that unit of exchange was used to purchase. PageRank does the opposite, assembling a carefully guarded, secret map of the network that is then interpreted into meaningful answers for users. That mirrored relationship hints at a broader set of changes that seem to be emerging as we incorporate algorithms more deeply into central machinery of human culture.

在此背景下，我们不仅可以将比特币解读为“可编程货币”意识形态的实现，更可以将其视为迈向可编程文化的巨变的寓言。比特币价值创造体系的核心原则，体现了文化生产各个领域类似的转变：日常事务看似相同，但多数投票、中央权威和验证结构正在被算法流程所取代。从高频交易到谷歌搜索，我们看到内在价值正在转化为一种网络价值，这种价值奖励情境、访问和互联互通。与此同时，对知识的追求也服务于对方法的追求，即对更优技术系统的追求。我们对Facebook、谷歌、苹果和其他科技公司黑匣子的日益增长的路径依赖，促使众多文化领域创作出针对这些系统进行优化的作品，创造出了类似于单一作物的文学作品，而这些作物对生态系统的重大变化的适应能力有限。

In this context, we can read Bitcoin not just as an implementation of the ideology of “programmable money,” but as an allegory for the sea change toward what we might call programmable culture. The central tenets of Bitcoin’s system for creating value epitomize similar shifts in all corners of cultural production, where daily business seems the same but the majority votes, central authorities, and validation structures are being subsumed by algorithmic processes. From HFT to Google Search, we see intrinsic value being transmuted into a kind of network value that rewards context, access, and interlinking. In parallel, the quest for knowledge is subservient to the quest for methods, for better technical systems. Our growing path dependency on the black boxes at Facebook, Google, Apple, and other tech companies pushes a huge range of cultural fields to produce work optimized for those systems, creating the belletristic equivalent of monoculture crops that have limited resilience to major changes in the ecosystem.

以新闻业为例。最近被赶下台的博客帝国Gawker Media的创始人尼克·登顿(Nick Denton)在2015年坦率地写道，他的公司犯下了种种错误：“我们——地球上最自由的记者——却成了Facebook算法的奴隶。” ³³登顿认为，他的公司已经忘记了发布引人入胜的新闻报道的核心使命，反而沉迷于吸引流量和注意力的算法游戏。如今，这个问题变得更加严重，因为像《纽约时报》这样的媒体公司允许Facebook直接托管他们的稿件，让用户更加沉迷于Facebook算法控制的内容流。正如科技记者艾德丽安·拉弗朗斯(Adrienne LaFrance)在《大西洋月刊》(The Atlantic)上所说，“Facebook是美国的新闻编辑”，为新闻网站带来了巨大且不断增长的流量^。34

Consider journalism. The recently dethroned founder of the Gawker Media blogging empire, Nick Denton, wrote candidly in 2015 about missteps at his company: “We—the freest journalists on the planet—were slaves to the Facebook algorithm.”³³ Denton was arguing that his company had lost sight of its central mission to break compelling news stories, growing obsessed with the algorithmic game of attracting traffic and attention instead. That problem has only gotten more serious now that media companies like the New York Times are allowing Facebook to host their copy directly, keeping users even more entrenched in a stream of content that Facebook’s algorithms control. As technology journalist Adrienne LaFrance argued in The Atlantic, “Facebook is America’s news editor,” driving a huge and growing percentage of the traffic to news sites.³⁴

2016 年，就在 Denton 哀叹 Facebook 对其记者的影响时，有消息称，这家社交网络却雇佣了自己的年轻记者团队，通过“热门话题”来挑选要推广的新闻。“热门话题”是一个引人注目的小工具，用于吸引用户关注当天的头条新闻。这些合同工被安排在一个闲置的会议室里，并被要求执行一种算法判断，根据一份名为“热门审查指南”的备忘录中精心设计的一套指令来挑选新闻。35^这些类似算法的指令和执行这些指令的人都被指责将自由主义偏见引入了“热门话题”，Facebook 此后宣布计划彻底改革“热门话题”的运作方式。这一发现强调了人们现在熟悉的观念，即人类隐藏在一个所谓的算法黑匣子里，但它也揭示了我们的文化系统已经变得多么唯我论。Facebook 的备忘录指示承包商只跟踪十个新闻网站，包括 CNN、BBC、Fox News 和《卫报》，以评估一个新闻是否值得“全国”关注。与此同时，所有这些网站的编辑都密切关注着 Facebook（记住，Facebook 并不是一个新闻机构），以了解哪些新闻能引起数字公众的关注。试图解读热门话题算法的新闻主管们，实际上是在窥视一面编辑趣味镜。36 ²⁰¹⁶年下半年，PayPal 联合创始人彼得·泰尔 (Peter Thiel) 支持第三方诉讼，导致 Gawker Media 破产，这再次体现了可编程文化和新的价值基本规则，并成为各大媒体的头条新闻。泰尔辩称，这“与其说是报复，不如说是特定的威慑”——他正确地理解了，作为一名算法亿万富翁，他有权审查一个他鄙视其淫秽报道的新闻机构^。37

In 2016 it emerged that even as Denton was lamenting the impact of Facebook on his journalists, the social network had hired its own cadre of young journalists to select stories for promotion through “Trending Topics,” a prominent widget for drawing user attention to top stories of the day. These contract workers were relegated to an unused conference room and asked to perform a kind of algorithmic judgment, selecting stories according to a carefully defined set of instructions laid out in a memo titled “Trending Review Guidelines.”³⁵ Both the algorithm-like instructions and the humans enacting them were accused of introducing liberal bias into Trending Topics, and Facebook has since announced plans to completely overhaul how Trending Topics functions. The revelation underscores the now-familiar idea of humans hiding within a supposedly algorithmic black box, but it also reveals how solipsistic our cultural systems have become. The Facebook memo instructed contractors to track just ten news sites, including CNN, the BBC, Fox News, and the Guardian, in evaluating whether a story merits “national” importance. Meanwhile, editors at all of those places closely monitored Facebook (which, remember, is not a news organization) to see what news was getting traction with the digital public. News directors trying to decipher the Trending Topics algorithm were really peering into an editorial funhouse mirror.³⁶ Later in 2016, yet another illustration of programmable culture and new ground rules for value erupted into the headlines when PayPal co-founder Peter Thiel drove Gawker Media into bankruptcy by supporting a third-party lawsuit. Thiel argued this was “less about revenge and more about specific deterrence”—and his correct interpretation that as an algorithmic billionaire he had the power to censor a news organization whose salacious coverage he despised.³⁷

衡量标准的巨大变化远不止于华尔街或新闻。文化成功的利害关系日益受到算法指标的驱动。在好莱坞特效工作室、竞选办公室、警察总部以及无数其他权力节点，能够充分利用集中计算来解决一系列抽象问题的玩家将赢得大奖。对于比特币来说，这意味着在区块链中添加一个新区块；对于其他领域，这可以是对其他相对透明、公开的竞赛（例如电影大片、选举、热门音乐单曲）的贡献——但实现这一目标的算法套利依赖于类似形式的数据驱动抽象、黑箱分析和密集处理。

The sea change in what counts does not stop with Wall Street or the news. The stakes of cultural success are increasingly driven by algorithmic metrics. In Hollywood special effects studios, at campaign offices, police headquarters, and in countless other nodes of power, the players who can best leverage concentrated computation to a set of abstracted problems win the prize. For Bitcoin this means adding a new block to the blockchain; for other fields it can be contributions to other relatively transparent, publicly accessible contests—blockbuster films, elections, hit musical singles—but the algorithmic arbitrage used to get there depends on similar forms of data-driven abstraction, black box analysis, and intensive processing.

计算主义使文化实践不仅可计算，而且可编程——易于进行集中编辑和修订。或许，这是我们逐渐从按照临时“每日酷站”模式组织的互联网，转变为拥有稳定层级和收入来源的互联网的必然结果。随着内容传播和过滤系统变得越来越精细、越来越复杂，生产内容所需的工具也变得更加自动化。像Gawker Media这样的公司，博主们炮制着大量短文，这些文章通常由搜索引擎流量和网络分析软件驱动，并会为那些迅速获得关注的文章提供额外奖励。由于社交媒体和更广泛的互联网的性质不断变化，他们写作首先是为了那些会决定是否在Facebook等网站上传播其作品的算法，其次才是为了那些随后会阅读并与他人分享的读者。

Computationalism makes cultural practices not just computational but programmable—susceptible to centralized editing and revision. Perhaps this is the inevitable consequence of our gradual shift from an Internet organized according to the ad hoc “cool site of the day” model to one with stable hierarchies and revenue streams. As the systems for disseminating and filtering content become more elaborate, more complex, the tools necessary for producing content have become more automated as well. Bloggers for companies like Gawker Media churn out short articles that are often driven by search engine traffic and web analytics software, with bonuses for pieces that gain rapid traction. Because of the changing nature of social media and the broader Internet, they are writing first for the algorithms that will choose whether or not to disseminate their work on sites like Facebook, and only second for the readers who will then read and share it with others.

丹顿的哀叹是一个例子，它表明，诸如新闻自由企业功能之类的文化价值观，可以通过修改分发内容和衡量成功的算法平台而逐渐改变。在流程机制下，价值的生产源于迭代：持续输出简短、主题鲜明的博客文章远比任何一篇博文的内容重要得多。算法对数据的应用胜过数据本身；信息的导航、合成和处理变得比简单地访问信息更重要。比特币协议巧妙地将这一原则编码在“挖矿”的概念中，明确地将验证交易的计算工作与新货币的生产联系起来。事实上，执行这一公共计算过程是创造更多比特币的唯一途径，这使得计算过程等同于政府铸币厂将纸片变成硬币的炼金术。神话中诺克斯堡式的“金山”“支撑”美元的，正是堆积如山的服务器，它们致力于在区块链上进行运算。

Denton’s lament serves as an example of how cultural values like the function of journalistic free enterprise can be gradually altered by revising the algorithmic platforms that distribute content and measure success. Under the regime of process, the production of value comes from iteration: the persistent output of short, topical blog posts is far more important than the content of any one of those posts. The application of algorithms to data trumps the data itself; the navigation, synthesis, and manipulation of information becomes more important than simply accessing it. The Bitcoin protocol elegantly encodes this principle in the notion of mining, explicitly linking the computational work of authenticating transactions with the production of new currency. Indeed, performing this public computational process is the only way to create more Bitcoins, making the process of computation equivalent to the alchemical magic of a government mint turning pieces of paper into specie. The mythological equivalent of a Fort Knox-style mountain of gold “backing” the dollar is the mountain of servers dedicated to churning through the blockchain.

在博主每天为 Gawker Media 等公司撰写十到十二篇文章的背景下，挖掘价值的方法各不相同，但其核心抽象概念却始终如一。这些博主所接触的在线广告收入系统创造了另一种处理方式，或者说“工作量证明”，即内容本身。博主撰写新文章后，会通过 Facebook 算法和用户传播，为 Facebook 和 Gawker Media 带来广告收入，有时博主本人也能从中分得一杯羹。这个过程将眼球和人们的注意力转化为金钱，而非创意。那些能够持续创作大量可点击内容的人将获得最大的回报。正因如此，博客的创作过程变得远比文章内容重要，而作者也因此沦为决定病毒式传播、分享率和人情味真正含义的算法的奴隶。

In the context of bloggers writing ten or twelve posts a day for companies like Gawker Media, the methods for mining value are different but the governing abstraction remains the same. The online ad revenue systems these bloggers interact with create another kind of processing or “proof of work,” the content itself. A blogger writes a new post that then gets circulated by Facebook algorithms and users, generating ad revenue for Facebook, for Gawker Media, and sometimes a narrow slice of that pie for the blogger. The process monetizes eyeballs and human attention, not ideas. Those who can consistently produce a high volume of clickable material are rewarded most greatly. This is how the process of blogging becomes much more important than the content of the posts, and how writers can become slaves to the algorithms that determine what virality, shareability, and human interest really mean.

价值从最终结果向过程的转移标志着一场漫长演变的顶峰，这场演变始于尤尔根·哈贝马斯著名的“资产阶级公共领域”。哈贝马斯认为，十八世纪有产中产阶级的出现（与《百科全书》的兴起相呼应）为公正的公共话语创造了一个新的空间，这种话语建立在资产阶级私人生活领域的真实价值之上。人们可以聚集在咖啡馆里阅读报纸，讨论当天的事件，践行公民参与，这种参与植根于他们对新兴现代资本主义体系的集体投入。作为有产者、公民和读者，他们可以在公共领域的受保护空间中忽略地位和个人利益，进行充满活力、批判性十足的对话。公共领域与自由媒体和稳定的政治讨论空间相互依存，从而能够对哈贝马斯所谓的“共同关切”领域达成共同理解和评价^。38

The migration of value from end result to process marks the culmination of a long evolution that began with what Jürgen Habermas famously called the bourgeois public sphere. Habermas argued that the emergence of a propertied middle class in the eighteenth century (parallel to the rise of the Encyclopédie) created a new space for disinterested public discourse founded on the truth values of the private, intimate sphere of bourgeois life. Men could gather in coffeehouses to read newspapers and discuss the events of the day, practicing civic engagement anchored in their collective investment in the emerging system of modern capitalism. As men of property, citizens, and readers, they could ignore status and personal interest in the protected space of the public sphere for the sake of a vibrant, critically energetic conversation. The public sphere emerged in codependence with a free press and a stable political space for discussion, allowing for the shared understanding and evaluation of what Habermas called areas of “common concern.”³⁸

资本主义与公共领域之间的密切联系，与可编程文化时代新闻业的加速演变形成了有益的对比。哈贝马斯理想化的公共领域，正是Gawker Media和Denton发现已在他们脚下消失的一部分。调查性新闻和独立媒体为哈贝马斯理想化的公民共同体提供了文化支架，提供了反映并推进读者共同价值观的思想和观点。虽然分析方法可能会不断发展，但公共领域的数据本身是可靠的、硬编码的：一套根植于国家和宗教机构、资本主义和咖啡馆的真理信念体系。公共领域从一开始就是一个备受争议的概念，因为它描绘了一幅关于谁可以参与这场资产阶级启蒙运动以及他们可以谈论什么的美好图景。但即使承认其缺陷，现代社会讨论和辩论的“集会”这一概念仍然具​​有批判性，尤其是在互联网似乎有望带来新的公众参与形式，这些形式与哈贝马斯历史阐释的关键特征相呼应的情况下。社交媒体、匿名论坛，甚至维基百科，有时似乎都能与启蒙运动时期的第一批公共办公室相媲美，为客观公正的批判性公共言论创造了空间。

The intimate connection between capitalism and the public sphere provides a useful contrast to the accelerating evolution of journalism in the era of programmable culture. Habermas’s idealized public sphere is part of what Gawker Media and Denton find to have disappeared beneath their feet. Investigative journalism and the independent press serve as the cultural scaffolding for the kind of civic community that Habermas idealizes, providing the ideas and opinions that reflect and advance the shared values of the reading public. While the methods of analysis might evolve, the data itself in this public sphere was dependable, hard-coded: a system of truth beliefs rooted in state and religious institutions, in capitalism and coffee shops. The public sphere was a contentious idea from the beginning because it painted a rosy picture about who got to participate in this bourgeois Enlightenment and what they got to talk about. But even accepting its flaws, the notion of a modern-day agora for societal discussion and debate has remained critically compelling, particularly as the Internet seems to promise new forms of public engagement that mirror the key features of Habermas’s historical interpretation. Social media, anonymous forums, and even Wikipedia seem at times to rival the first public offices of the Enlightenment, creating a space for disinterested, critical public speech.

丹顿的陨落是一个警示故事，它表明了原本构建公正公共话语的框架正在如何演变成一种截然不同的结构。在可编程文化时代，宣传和话语伦理可以像比特币一样互换，从根本上改变公共领域赖以生存的社会契约。博主们在保护消息来源或承认利益冲突方面遵循哪些规则？在Facebook上，《纽约时报》和《华盛顿邮报》的监察员隐藏在算法不透明的把关之下。蒂尔认为，他对Gawker的怨恨源于他们滥用新闻业的名义欺凌和羞辱个人。对他而言，财产所有权的伦理以及由此衍生的隐私高于任何公共领域的概念。

Denton’s fall is a cautionary tale illustrating how that scaffolding for disinterested public discourse is turning into a very different sort of structure. In the era of programmable culture, publicity and the ethics of discourse can be as fungible as Bitcoin, fundamentally changing the social contract on which the public sphere depends. What rules do bloggers follow about protecting sources, or acknowledging conflicts of interest? On Facebook, the ombudspeople of the New York Times and the Washington Post disappear behind the opaque gatekeeping of algorithms. Thiel argues that his grudge against Gawker originated in their abuse of the mantle of journalism to bully and humiliate individuals. For him, the ethics of property ownership and a privacy derived from that ownership trumped any notion of a public sphere.

可编程文化彻底颠覆了公共领域：过去构成“共同关注”的文化数据日益私有化。普通公民的私生活公开化，数据经纪人、社交媒体公司和其他机构可以通过商业手段获取用户私生活的访问权。他们以访问新近私有化的在线社区（例如Facebook和Twitter）的权限交换用户私生活的访问权。算法精英成员逐渐淡出公众视野，其中一些人呼应了泰尔的自由主义观点：“我不再相信自由与民主可以兼容。” ³⁹金融曾是资产阶级资本主义中最私密的记录，但比特币将其变成了金钱的公共领域。通过将金融交易置于玻璃盒子中，比特币的创造者颠覆了公共领域的审查模式，允许公众审查商业交易，同时保持文化立场和身份的私密性。共识是围绕金融交易而不是批判性判断而形成的，并且金钱成为政治言论的唯一真正形式，以延伸美国最高法院在《联合公民》中的逻辑。

Programmable culture turns the public sphere inside out: the cultural data that used to make up the arena of “common concern” is increasingly privatized. The private life of the average citizen becomes public, commercially accessible to data brokers, social media companies, and others who trade access to the newly privatized sphere of online community (through sites like Facebook and Twitter) in exchange for commercial access to users’ private lives. Members of the algorithmic elite retreat from public view, some of them echoing Thiel’s libertarian sentiment: “I no longer believe that freedom and democracy are compatible.”³⁹ Finance was the most private of registers in bourgeois capitalism, but Bitcoin turns it into a public sphere for money. By placing financial transactions into the glass box, the creators of Bitcoin invert the public sphere’s model of scrutiny, allowing public inspection of commercial exchange but keeping cultural positions and identities private. Consensus gathers around financial transactions rather than critical judgments, and money becomes, to extend the logic of the U.S. Supreme Court in Citizens United, the only real form of political speech.

这种新的金融公共领域在其反乌托邦的极端情况下，将银行账户变成了公民：超级政治行动委员会（SuperPAC）和风险投资家就当今的重大事件进行重要对话，并由那些足够富有、能够发声并被倾听的个人参与其中：不仅有蒂尔，还有沃伦·巴菲特、比尔·盖茨和杰夫·贝佐斯。从更积极的一面来看，我们可以看到金钱的公共领域通过Kickstarter和Indiegogo等筹款网站改变艺术，这些网站允许通过众筹（或某种财务投票）集体批准新项目。众筹的算法流程会奖励那些掌握私有化宣传方法的人：精彩的介绍视频、频繁的更新、分级奖励结构以及有效利用社交媒体来提高知名度。

At its dystopian extreme, this new financial public sphere makes bank accounts into citizens: SuperPACs and venture capitalists have the important conversations about events of the day, joined by those individuals wealthy enough to speak and be heard: not just Thiel but Warren Buffett, Bill Gates, and Jeff Bezos, for example. On a more positive note, we can see the public sphere of cash transforming the arts through fundraising sites like Kickstarter and Indiegogo, which allow for the collective approval of new projects through crowdfunding, or a kind of financial voting. The algorithmic process of crowdfunding rewards those who master the methods of privatized publicity: a strong introductory video, frequent updates, tiered reward structures, and effective use of social media to raise awareness.

从这个角度来看，丹顿的文章是对金钱公共领域诱惑的猛烈抨击，因为它正在破坏新闻业的公共领域。当博主和记者被赋予受众指标和目标时，他们正在努力应对一种新的公众关注模式，这种模式以金钱计算，但实际上衡量它的却是CPU周期、链接、悬停和点击次数。旧的新闻业在报纸销售业务和思想市场之间插入了一套启蒙运动的抽象概念——第四等级的概念，即服务于公众利益的自由独立媒体。算法时代的流程评估消除了这些中介层，使作家和编辑能够实时了解市场对其作品的反应。直接参与这种反馈循环可以导致标题和报道角度的改变，彻底探索新的主题，并不断适应社交媒体算法的规范效应（就Gawker而言，还包括第11章破产）。

Read in this light, Denton’s post is a diatribe against the seductions of the public sphere of money as it undermines the public sphere of journalism. When bloggers and reporters are given audience metrics and goals, they are grappling with a new model of public attention denominated in dollars, but really measured in CPU cycles, links, hovers, and clicks. The old journalistic enterprise had interposed a set of Enlightenment abstractions—the notion of the fourth estate, a free and independent press serving the public good—between the business of selling newspapers and the marketplace for ideas. The valuation of process in the age of algorithms deletes those mediating layers, allowing writers and editors to see the market response to their work in real time. Direct engagement with that feedback loop can lead to changing headlines and angles for stories, pursuing new topics entirely, and continually adapting to the normative effects of social media algorithms (and, in Gawker’s case, to Chapter 11 bankruptcy).

哈贝马斯迅速指出，随着大众媒体和现代广告的兴起，资产阶级公共领域逐渐消亡，但客观公正的公共话语的理想愿景依然存在于记者、社会企业家、维基百科人以及其他许多人的愿望中，他们将互联网视为围绕“共同关切”进行集体参与的新平台。向过程价值的转变也改变了我们脚下的认识论基础。尽管蒂尔在Gawker诉讼案中采取了经典的资本主义式干预，但显而易见的是，那些植根于资本主义、个人财产以及最终源于启蒙运动本身的共享客观真理，已经让位于一种建立在可互换数据和过程真理之上的新客观性——我们或许可以愤世嫉俗地称之为“文化即服务”。随着我们将越来越多的生活、公共和私人对话、信仰和信念投入到算法文化机器中，我们也认同了真理存在于分析、抽象和数据挖掘之中的理念。

Habermas was quick to point out that the bourgeois public sphere faded away with the rise of mass media and modern advertising, but the ideal vision of disinterested public discourse persists in the aspirations of journalists, social entrepreneurs, Wikipedians, and many others who see the Internet as a new platform for collective engagement around “common concerns.” The shift to processual value has also shifted the epistemological floor beneath our feet. Despite Thiel’s classically capitalistic intervention in the Gawker lawsuit, it is clear that shared, objective truths grounded in capitalism, personal property, and ultimately the Enlightenment itself have given way to a new objectivity founded on fungible data and processual truths—what we might cynically call culture as a service. As we invest more of our lives, our public and private conversations, our faith and beliefs, in algorithmic culture machines, we invest in the idea that truth resides in analytics, abstraction, and data-mining.

挖矿价值

Mining Value

数据处理的价值化对文化生产有着深刻的影响。作为文化消费者，我们现在通过文本之间的联系而非其内容来评价文本的意义。批评家不再主要因为缩小范围、消除不良链接和树立排他性的知识品味而受到赞扬。像保琳·凯尔或哈罗德·布鲁姆这样的人物，他们的职业生涯建立在告诉我们什么该忽略、什么该赞扬的基础上，而现在他们被提供持续包容性网络文化服务（或流程）的策展人所取代。例如，玛丽亚·波波娃的《脑力精选》策展精妙，提供了一场范围广泛的、连续的知识盛宴。它作为波波娃“痴迷地”维护的持续内容流的地位，使这条内容流比她引用的单个作品更有价值。40^她的创作过程的表现，即《脑力精选》这个复杂的人机文化机器，成为了她读者感兴趣的对象。她以自己的方式，从一种我们常说的“策展”式的审美套利中挖掘价值。她将这些碎片组合起来，以一种吸引眼球的方式，将不同的元素转化为引人入胜的全新体验。优秀的博主也同样如此，他们为那些以持续、及时地制作媒体流而闻名的网站收集有价值的内容。

The valorization of data processing has provocative implications for cultural production. As cultural consumers, we now evaluate the meaning of text through its connections more than its substance. The critic is no longer celebrated primarily for narrowing the field, eliminating bad links, and performing an exclusive model of intellectual taste. Figures like Pauline Kael or Harold Bloom who built their careers telling us what to ignore as much as what to celebrate are being replaced by curators who provide an ongoing service (or process) of inclusive, networked culture. The inspired curation of Maria Popova’s Brain Pickings, for example, offers a wide-ranging, serial feast of intellectual pleasures. Its status as a persistent stream of content, one that Popova “obsessively” maintains, makes the stream more valuable than the individual works she cites.⁴⁰ The performance of her creative process, the complex human-computational culture machine that is Brain Pickings, becomes the object of interest for her readers. She is, in her own way, mining value in a particular kind of aesthetic arbitrage that we often call curation. She assembles these pieces in a way that captures attention, that transmutes disparate elements into a compelling new experience. Great bloggers do the same, gleaning nuggets of content for sites that are celebrated primarily for their consistent, timely production of a media stream.

换句话说，驱动比特币的同样的过程价值逻辑也在重塑其他文化形式。我们正在学习或受到那些将过程提升到价值地位的算法和计算系统的影响。我们正在慢慢地用永恒丰富的文化、类似Netflix的自助餐——融合了多种媒体、副文本和知识框架——取代人类特有的线性关注的法典模型——取代我们特有的文化。这更像是一朵思想云，而非意识流。策划这些思想云需要像波波娃这样的人优雅地运用由计算内存、人类贡献者和读者网络以及各种协作平台组成的混合算法机器。但正如文化生产的主要领域正在适应过程价值逻辑一样，我们，作为这些产品的受众，也在改变我们的价值观以及作为其作品消费者的习惯。

In other words, the same processual logic of value that drives Bitcoin is remaking other cultural forms as well. We are learning from, or being influenced by, algorithms and computational systems that elevate process to the status of value. We are slowly replacing the very human codex model of linear attention with an eternal abundance of culture, a Netflix-like smorgasbord that integrates multiple media, paratexts, and intellectual frameworks—a cloud of ideas rather than a stream of consciousness. Curating these clouds requires human beings like Popova to work gracefully with hybrid algorithmic machines consisting of computational memory, networks of human contributors and readers, and various collaborative platforms. But just as the major fields of cultural production are adapting to the logic of processual value, we, the audience of these products, are also changing our values along with our habits as consumers of their work.

想想我们现在在袖珍电脑里随手就能找到的众多形式的文化信息：所有能想到的消费品的即时比较销售排名，以及相关的热门评论；所有能想到的医学话题、育儿问题或成瘾治疗方案的坦诚公开讨论；以及对已知世界中每家餐馆、咖啡馆和博物馆的古怪而主观的评论。所有这些信息都无法像法国社会学家皮埃尔·布迪厄在前算法时代所描绘的那样，通过传统的、高度规范的文化生产渠道持续大量地获得（或者，通常甚至以有限的形式获得）。41^这些信息几乎全部是私人知识，在“大反转”之前被刻意屏蔽在哈贝马斯的公共领域之外。在20世纪的大部分时间里，如果一位女性想研究避孕方法及其副作用，或者一位男性想阅读关于同性恋出柜的叙述，根本没有搜索引擎，也没有公开的、受认可的信息来源。一位想更换淋浴龙头阀杆的房主，或者一位想研究新型造型涂料的爱好者，除了当地社区团体、商店以及（某些情况下）图书馆之外，几乎没有其他资源可以利用。在互联网出现之前，这些都是私人事务，而非公共事务。

Consider the many forms of cultural information we now casually expect to find in our pocket computers: instant comparative sales rankings of every imaginable consumer good, plus popular reviews thereof; frank, public discussions of every conceivable medical topic, childrearing problem, or addiction protocol; quirky and subjective reviews of every eatery, coffee shop, and museum in the known world. None of this information was available in such persistent abundance (or, often, in even limited form) through the traditional, highly normative avenues of cultural production as French sociologist Pierre Bourdieu mapped them in the pre-algorithmic era.⁴¹ It was almost all private knowledge, deliberately shielded from the Habermasian public sphere before the great inversion. For much of the twentieth century, if a woman wanted to research birth control options and side-effects, or a man wanted to read narratives about coming out as gay, there was no search engine, no public, sanctioned source of information at all. A homeowner who wanted to replace a shower faucet stem valve or a hobbyist researching a new kind of modeling paint would have few resources at their disposal beyond local community groups, stores, and, for some questions, libraries. Before the Internet, these were private concerns, not common concerns.

这种转变如此深刻，如此明显，以至于即使对于我们这些经历过其近期阶段的人来说，也已变得难以察觉。构成网络深层结构的结构收集并存储着人类的注意力、亲密关系、记忆以及我们所有其他的数字痕迹。为了利用这些丰富的信息，它们都利用了PageRank之类的算法套利系统，将我们引导至特定的论坛，特定的文章，以满足某些特定的需求。对我们许多人来说，关于生活的每一个问题都始于对搜索引擎或其他算法数据库的查询，这种新的可读性以一种深刻的认识论方式改变了我们与现实的关系。我们欣慰地知道，以后随时可以谷歌搜索，因此逐渐接受了信息套利比信息本身更重要这一事实。注释和评论已经完全取代了文章和书籍，以至于作家卡尔·塔罗·格林菲尔德在《纽约时报》上撰文指出，我们集体倾向于基于副文本和元数据伪造文化素养：

This transformation is so profound, so obvious, that is has become invisible even to those of us who have lived through its recent phases. The structures that constitute the deep fabric of the web collect and store human attention, intimacy, memories, and all our other digital traces. To make use of this abundance, they all leverage algorithmic arbitrage systems like PageRank, funneling us toward the particular forum, the very specific article, that addresses some particular need. For many of us, every single question about our lives starts with a query to a search engine or another algorithmic database, and this new legibility changes our relationship to reality in a deep, epistemological way. Comforted in the knowledge that we can always Google it later, we have gradually accepted that the arbitrage of information is more significant than the information itself. The gloss and the comment have overtaken the article and the book so completely that author Karl Taro Greenfield wrote in the New York Times about our collective tendency to fake cultural literacy on the basis of paratext and metadata:

假装无所不知从未如此容易。我们从Facebook、Twitter或邮件新闻提醒中挑选热门、相关的信息，然后反复阅读。……在海量数据泛滥的今天，对我们来说，重要的并非亲身体验过这些内容，而是仅仅知道它们存在——并且对它们有自己的看法，能够参与其中^。42

It’s never been so easy to pretend to know so much without actually knowing anything. We pick topical, relevant bits from Facebook, Twitter or emailed news alerts, and then regurgitate them. … What matters to us, awash in petabytes of data, is not necessarily having actually consumed this content firsthand but simply knowing that it exists—and having a position on it, being able to engage in the chatter about it.⁴²

就像高频交易员、谷歌的PageRank、比特币一样，我们不再关心信息包的细节，只要我们能够有效地套利，在数据失去信誉之前将其转化为优势即可。文化消费的奇观，即处理、评判和分享的公共行为，正是新的文化经济。

Like the HFT traders, like Google’s PageRank, like Bitcoin, we no longer care about the details of our information packets so long as we can effectively arbitrage them, transferring the data to our advantage before it loses cachet. The spectacle of cultural consumption, the public act of processing, judging, and sharing, is the new cultural economy.

解读系统本身比解读系统构建之初旨在传递的内容更为重要。在最激烈的环境下，这些注意力架构会创造出自身持续的信念流，某种文化价值区块链。像维基百科这样的网站，甚至社交媒体上的公开对话，都提供了详细的交易流，允许参与者根据贡献历史来核对某一辩论线索（但有一个显著的例外：除了维基百科之外，几乎所有这类平台都允许以隐形的方式抹去这些痕迹）。与比特币一样，注意力的公共账本完全透明，但生成它所涉及的人工计算和算法计算却被掩盖。维基百科贡献者或 Facebook 评论者的动机可能难以辨别，但平台本身却呈现出一种扁平化、民主化、易于访问的形象^。43

Reading the system has become more important than reading the content the system was built to deliver. At their most intense, these architectures of attention create their own persistent streams of belief, a kind of cultural value blockchain. Sites like Wikipedia and even public conversations on social media offer a detailed transaction stream allowing participants to check a thread of debate against a history of contributions (with the notable exception that almost all of these platforms, aside from Wikipedia, allow the invisible removal of these traces). Like Bitcoin, the public ledger of attention is completely transparent, but the human and algorithmic computation involved in generating it is obscured. The motivations of Wikipedia contributors or Facebook commenters can be difficult to discern, but the platform itself presents an appearance of flat, democratic accessibility.⁴³

这种公共处理，这种类似区块链的文化话语生成，其诱惑力恰恰在于其可见性，以及其令人激动的透明度。文化处理（无论是比特币交易、Facebook点赞、点击驱动的政治行动，还是维基百科更新）本身就成为一种奇观，将我们从系统不可见的层面转移开来，如同玻璃盒子里上演的魔术。其背后的真相使其更具吸引力：比特币、维基百科，甚至社交媒体新闻推送都拥有某种乌托邦式的合法性。它们确实有可能绕过传统的、偶尔专制的信息控制结构，从阿拉伯之春到快速汇编关于重大恐怖袭击的详细客观的百科全书条目。这种奇观是协作的、变革性的、实时的。这种奇观承诺通过无处不在的计算，使整个世界变得清晰易读、易于改变。可编程文化承诺，只需轻触图标，即可实现民主、问责和清晰度。

The seduction of this public processing, the blockchain-like generation of cultural discourse, lies precisely in its visibility, its thrilling transparency. Cultural processing (whether of Bitcoin transactions, Facebook likes, click-driven political activism, or Wikipedia updates) becomes its own spectacle, distracting us from the invisible sides of the system, like a magic trick performed in a glass box. The seduction is all the more powerful because of the truth behind it: Bitcoin, Wikipedia, and even the social media newsfeeds have some utopian legitimacy. They do really have the potential to route around traditional, occasionally despotic structures for controlling information, from the Arab Spring to the rapid assembly of detailed, objective encyclopedia entries on major terrorist attacks. The spectacle is collaborative, transformative, live. The spectacle promises to make the whole world legible and mutable through the sheer ubiquity of computation. Programmable culture promises democracy, accountability, and clarity at the tap of an icon.

当然，挖掘价值的算法概念依赖于一种任意的行为语法。44比特币对其挖矿集体施加的工作量证明在其他系统中也有类似之处，例如博主为了适应 F​​acebook 算法而优化他们的故事^。虽然这些文化和计算劳动的形式通常以人类的标准来看是任意的——制造工作是为了过滤掉系统中杂乱的边缘——但它们在规范原本杂乱无章的社区方面，围绕既定的算法目标，满足了重要的需求。比特币矿工们围绕着他们的 CPU 周期团结起来，通过一组既透明又隐秘的计算卡特尔推动着系统的发展。仪式化地使用“点赞”和“好友”按钮是 Facebook 的工作量证明，鼓励我们参与进来，让我们的贡献更有价值……而在这套语法之下，隐藏着该平台的第二语言——广告和商业化用户数据。对于 PageRank 来说，工作量证明是最引人注目的：它构建了定义互联网本身的链接和引用。这些系统最终的魅力在于，通过我们自身的贡献，重构和扩展信息链，形成一个不断强化的反馈循环。早在 20 世纪 90 年代初，信息学者菲利普·阿格雷就预测了这种可编程性（他称之为“信息捕获”）的许多后果：

But of course the algorithmic notion of mining for value depends on an arbitrary grammar of action.⁴⁴ The proof of work imposed by Bitcoin on its mining collectives has its analogs in other systems as well, like bloggers optimizing their stories for Facebook algorithms. And while these forms of cultural and computational labor are often arbitrary by human standards—make-work designed to filter out the messy fringes of the system—they serve important needs in normalizing otherwise haphazard communities around set algorithmic objectives. Bitcoin miners unite around their CPU cycles, propelling the system forward with a set of computational cartels that are both transparent and occult. The ritualistic use of “like” and “friend” buttons are Facebook’s proof of work, encouraging us to engage to make our contributions count … and beneath this grammar lies the platform’s second language of advertisements and commercialized user data. For PageRank, the proof of work is the most compelling: the construction of the links and references that define the Internet itself. The ultimate seductive power in each of these systems lies in the reinforcing feedback loop of refactoring and extending the chain through our own contributions. In the early 1990s, information scholar Philip Agre predicted many of the consequences of this kind of programmability, which he called information capture:

抛开所有学科的孤立性，也无论人们对这一过程的自觉理解如何，构建这些本体论的实践本身就是一种社会理论化的形式。这种理论化并非仅仅是学术活动，而是一个更宏大的物质过程的一部分，通过这一过程，这些新的社会本体论在某种特定意义上得以实现^。45

The practice of formulating these ontologies is, all disciplinary insularity aside, and regardless of anyone’s conscious understandings of the process, a form of social theorization. And this theorizing is not simply a scholastic exercise but is part of a much larger material process through which these new social ontologies, in a certain specific sense, become real.⁴⁵

我们的参与使系统更加完善，并不断训练我们掌握流程驱动估值的精髓。

Our participation wears grooves deeper into the system and continually trains us in the fine arts of process-driven valuation.

用可编程领域取代公共领域，最终是用一种阅读形式取代另一种：一种新的惯用框架，用来构建世界。离开十八世纪伦敦咖啡馆里的《旁观者》读者，我们发现算法过程的读者正在星巴克的Wi-Fi上解读推特信息和网络流量。新算法领域的语法掩盖了某些操作，同时又使其他操作更加清晰可见，而区块链的奇观仅仅是我们赋予这些文化系统巨大力量的新方式之一。我们都越来越擅长批判性套利的实践，因为我们的系统，泛滥的文化数据流，需要它。玻璃盒子和黑盒子的相互作用要求对数据过程进行算法式的关注，这是一种我们都已在实践的算法式解读。

Replacing the public sphere with the programmable sphere is ultimately a substitution of one form of reading for another: a new idiomatic frame to put on the world. Leaving behind the readers of the Spectator in an eighteenth-century London coffeehouse, we find the readers of algorithmic process, interpreting Twitter feeds and web traffic counts over Wi-Fi at Starbucks. The grammar of the new algorithmic sphere obscures certain operations while making others more visible, and the spectacle of the blockchain is merely one of the newer ways that we have invested these cultural systems with enormous power. We are all becoming more adept at the practice of critical arbitrage because our systems, the flooded streams of cultural data, require it. The interplay of glass boxes and black boxes demands an algorithmic attention to data processes, an algorithmic reading that we are all practicing already.

笔记

Notes

机器学习

Machine Learning

算法想象力究竟意味着什么？在本文的尾声，我想探讨这样一个世界的含义：文化机器正在承担越来越多批判性和创造性的工作，而这些工作曾经是人类独有的、内在的。

What does it mean to talk of an algorithmic imagination? In this coda, I want to explore the implications of a world where culture machines are taking on a growing share of the critical and creative work that used to be distinctively, intrinsically human.

谷歌的 DeepMind 跨过了这样一个门槛，工程师们让它识别和增强循环播放的图像中的各种特征，创作了一系列达利或博斯式的画作，其中算法机器智能将阴影和交叉点转化为眼睛、脸庞和其他可识别的视觉元素。最终的作品展现出一种独特的、陌生的视角，艺术家的眼睛似乎能够想象并实现我们从照片中永远无法预料到的东西。正如数学史学家大卫·柏林斯基所说，这种“外星智慧”的景象源于一种强大且富有想象力的新型算法生产形式。它也让我们得以一窥整个过程，一系列富有想象力的实践，它们以一种与我们讲述的人工智能故事（例如《2001：太空漫游》中的 HAL或《她》中的萨曼莎）截然不同的方式唤起艺术意向性。至少，我们认为我们在程序生成的奇怪渲染中看到了这一瞥——也许这只是人类观察者代表机器想象出这些电子羊，遵循我们持续的冲动，将拟人化并将意向性投射到我们遇到的每一个复杂系统中。

Google’s DeepMind crossed one such threshold when engineers tasked it with identifying and enhancing various image features on a repeating loop, creating a series of Dalí- or Bosch-like pictures where an algorithmic machine intelligence turned shadows and intersections into eyes, faces, and other recognizable visual elements. The resulting work shows the signs of a distinct, alien perspective, an artist’s eye that seems to imagine and then realize things that we would never have expected from a photograph. That glimpse of “intelligence on alien shores,” as mathematical historian David Berlinski puts it, emerges from a newly powerful and perhaps imaginative form of algorithmic production. It also offers a glimpse of process, a set of imaginative practices, that evoke artistic intentionality in a way that feels very different from the stories we tell about artificial intelligence, like HAL from 2001 or Samantha from Her. At least, we think we see that glimpse in the strange renderings the program produces—perhaps it is just the human observers dreaming up these electric sheep on behalf of the machine, obeying our persistent impulse to anthropomorphize and project intentionality into every complex system we encounter.

DeepMind 因其成就的广泛性而引人注目。在谷歌收购 DeepMind 的几周前，该公司因一款机器学习算法而登上国际新闻，该算法在无人监督的情况下，学会了比普通人更胜一筹地玩 29 款雅达利游戏。1如今^，同样的算法已经取代了谷歌“60 个基于规则的手工系统”，涵盖从图像识别到语音转录等各个领域。2^最引人注目的是，2016 年 3 月，DeepMind 的 AlphaGo 以 4-1 击败围棋大师李世石，证明了其对人类最精妙、最具艺术性的游戏之一的征服。3^在经历了漫长的低迷之后，谷歌和其他一系列研究机构似乎在开发能够优雅地适应各种概念挑战的系统方面取得了进展。这一阶段性的转变催生了一批新的中心和项目，它们致力于应对人工智能的潜在后果，并将哲学家、技术专家和硅谷亿万富翁团结起来，共同探讨一个问题：一台真正具有思考能力的机器是否会对人类构成生存威胁。

DeepMind is remarkable for the range of its achievements. A few weeks before Google purchased it, the company made international news with a machine learning algorithm that had learned to play twenty-nine Atari games better than the average human with no direct supervision.¹ Now the same algorithm has replaced “sixty handcrafted rule-based systems” at Google, from image recognition to speech transcription.² Most spectacularly, in March 2016 DeepMind’s AlphaGo defeated go grandmaster Lee Sedol 4–1, demonstrating its conquest of one of humanity’s subtlest and most artistic games.³ After a long doldrums, Google and a range of other research outfits seem to be making progress on systems that can gracefully adapt themselves to a wide range of conceptual challenges. This phase shift has produced a new crop of centers and initiatives grappling with the potential consequences of artificial intelligence, uniting philosophers, technologists, and Silicon Valley billionaires around the question of whether a truly thinking machine could pose an existential threat to humanity.

在描述图灵测试的论文中，艾伦·图灵还探讨了机器智能这个更广泛的问题：意识算法。图灵测试在很多方面都证明了建立智能指标的荒谬性；我们所能做的最好的事情就是进行对话，看看机器模拟人类的效果如何。但是，图灵提出，如果我们真的实现了这样的突破，那么考虑“儿童机器”的概念将非常重要，因为它可以学习我们想要教的东西。4^这种哲学立场支撑着 DeepMind 和许多其他近期算法智能的突破，这些突破都源自目前炙手可热的计算机科学子领域——机器学习。事实上，某些版本的机器学习支撑着本书中提到的许多算法：Siri、Netflix，当然还有谷歌，都依赖这些系统来解析复杂数据并做出决策。

In the paper where he described the Turing test, Alan Turing also took on the broader question of machine intelligence: an algorithm for consciousness. The Turing test was in many ways a demonstration of the absurdity of establishing a metric for intelligence; the best we can do is have a conversation and see how effective a machine is at emulating a human. But, Turing proposed, if we do achieve such a breakthrough, it will be important to consider the concept of the “child machine,” which learns what we wish to teach.⁴ That philosophical position underpins DeepMind and many other recent algorithmic intelligence breakthroughs, which have emerged from the currently incandescent computer science subfield of machine learning. Some version of machine learning, in fact, underpins many of the algorithms at work in this book: Siri, Netflix, and of course Google all depend on these systems to parse complex data and make decisions.

机器学习可以基于一些基本结构进行运作，例如神经网络、贝叶斯分析或进化适应。通常，最复杂的机器学习系统会融合多种方法的某些方面。学习算法一旦构建完成，就可以进行训练，最好是在能够提供大量期望结果或问题示例的海量数据上进行训练。学习算法会根据某种相对成功的信号（例如，衡量算法与正确猜测邮政地址模糊图像上数字的接近程度）随时间迭代。给定时间、数据和对问题的精确描述，机器学习算法可以为该问题创建稳健的解决方案。

Machine learning can operate according to a few basic structures, such as neural networks, Bayesian analysis, or evolutionary adaptation. Often the most sophisticated machine learning systems combine aspects of multiple approaches. Once a learning algorithm has been constructed, it can be trained, preferably over a huge corpus of data that can provide many examples of the desired outcome or problem. The learning algorithm iterates over time based on some kind of signal of relative success (e.g., measuring how close the algorithm got to correctly guessing the numbers on a blurry image of a postal address). Given time, data, and a precise statement of the problem, a machine learning algorithm can create a robust solution to that problem.

机器学习与本书如此相关的原因是，这个问题的“解决方案”实际上就是一种算法本身。正如机器学习研究员 Pedro Domingos 在《终极算法》一书中所言：

What makes machine learning so relevant to this book is that the “solution” to that problem is in fact an algorithm itself. As machine learning researcher Pedro Domingos puts it in The Master Algorithm:

每个算法都有输入和输出：数据输入计算机，算法对其进行处理，然后输出结果。机器学习则将这一过程反过来：输入数据和期望结果，然后输出将两者相互转化的算法。学习算法（也称为学习器）是能够生成其他算法的算法^。5

Every algorithm has an input and output: the data goes into the computer, the algorithm does what it will with it, and out comes the result. Machine learning turns this around: in goes the data and the desired result and out comes the algorithm that turns one into the other. Learning algorithms—also known as learners—are algorithms that make other algorithms.⁵

这些算法产生的过程、中介和跨越差距的实现将工程师的问题和解决方案统一起来。而且，由于它们通过模拟神经网络的百万次迭代来实现这一点，它们产生的计算结构往往既有效又难以捉摸。当 Netflix 的 Todd Yellin 谈到“机器中的幽灵”或数学家 Steven Strogatz 将自动化科学描述为“洞察力的终结”时，机器学习解决方案正是他们所想的。6^问题实际上是规模问题。作为人类，我们可以构思和管理庞大的数据集，但我们对这些数据集提出有趣问题的能力是有限的。我们可以自动化一些提问，但我们也会失去背景，因此答案可能是正确的，但没有提供任何信息。

These algorithms produce the process, the mediation, the gap-spanning implementation that unites their engineers’ problems and solutions. And, since they do this through the million-fold iteration of simulated neural networks, the computational constructs they spit out tend to be both effective and inscrutable. When Netflix’s Todd Yellin talks about the “ghost in the machine” or mathematician Steven Strogatz describes automated science as the “end of insight,” machine learning solutions are precisely what they have in mind.⁶ The problem is effectively one of scale. As humans we can conceive of and manage vast data sets, but our ability to ask interesting questions of those data sets is limited. We can automate some of the question-asking but we also lose context, so the answers may be true but uninformative.

换句话说，机器学习为该项目结束时计算主义的回归提供了完美的陪衬：计算宇宙的逻辑可以用来解开宇宙本身的奥秘。通过组装遵循一些简单计算定律的系统，我们可以迭代出高度复杂的解决方案，以解决那些难以用更直接（例如，人为设计的）算法方法解决的问题。那些从混沌中闪现的复杂秩序和过程，暗示着某种想象力在起作用，并表明定义机器学习的开放式迭代本身可能就是一种有限的想象力实践。而我们对这个问题的理解，以及我们进入计算海洋的途径，一如既往地依赖于隐喻。

In other words, machine learning provides a perfect foil for the return of computationalism at the end of this project: the ways that the logic of a computational universe could be used to unlock the mysteries of that very universe. By assembling systems that follow a few simple computational laws, we can iterate toward highly sophisticated solutions to difficult problems that resist more straightforward (e.g., human-designed) algorithmic approaches. Those flashes of complex order and process emerging from chaos hint at a kind of imagination at work, and suggest that the open-ended iteration that defines machine learning might be a limited form of imaginative practice in its own right. And our purchase on that question, our access to the ocean of computation, depends as always on metaphor.

算法想象力

Algorithmic Imagination

斯坦尼斯拉夫·莱姆在其杰作《索拉里斯星》中设想了一支科学探险队，他们探索一个神秘的海洋星球，那里永不停歇的海浪抛射出令人着迷的形状和结构。7人类科学家不断寻求解释这些海浪，以便与这个可能的外星智慧生物进行交流。^一派研究人员认为，索拉里斯星的海洋本身就是一个巨大的思维，一个在探险队眼前进行计算的复杂系统。事实上，我们可以将莱姆的这部小说解读为一种幻想性错觉的寓言，即对意义和模式的无休止追寻，而这正是启蒙运动的核心。莱姆似乎在暗示，工具理性的终极驱动力，即为了寻求答案而不断审视世界，可能只是疯狂：我们为了构建关于真理的故事而将幻想投射到这个世界，就像小说中的人物必须与从他们的过去投射出来的形象搏斗一样。

In his remarkable novel Solaris, Stanislaw Lem imagines a scientific expedition confronting a mysterious ocean planet where the perpetual motion of the waves throws up fascinating forms and structures.⁷ The human scientists continually seek to interpret them, to communicate with this possible alien intelligence. One school of researchers proposes that the ocean of Solaris itself is a giant mind, a complex system working through its calculations before the expedition’s eyes. Indeed, it’s possible to read Lem’s novel as an allegory for the apophenia, the endless hunt for meaning and patterns, that lies at the heart of the Enlightenment project. Lem seems to hint that the ultimate drive of instrumental reason, of continually interrogating our world in the quest for answers, may only be madness: fantasies that we project onto the world in order to construct a story about truth, just as the characters in the novel must grapple with figures projected from their pasts.

Solaris 上海洋无休止的象征性翻腾，为当今算法想象力的概念提供了一个颇具启发性的隐喻，尤其是在计算学家的框架下。谷歌机器学习团队负责人、奇点辩论的关键人物雷·库兹韦尔 (Ray Kurzweil) 在他的纪录片《超越人》(Transcendent Man)中呼应了这一观点：

The endless symbolic churning of the oceans on Solaris offers a provocative metaphor for the notion of algorithmic imagination today, particularly as it’s framed by computationalists. Ray Kurzweil, leader of Google’s machine learning group and a key figure in the singularity debates, echoes this point in his documentary Transcendent Man:

我在想，海洋代表了多少计算。我的意思是，所有水分子都在相互作用。这就是计算。它非常美丽。我一直觉得它很抚慰人心。而这正是计算的意义所在。捕捉我们意识中这些超越的瞬间^。8

Well I was thinking about how much computation is represented by the ocean. I mean it’s all these water molecules interacting with each other. That’s computation. It’s quite beautiful. And I’ve always found it very soothing. And that’s really what computation’s all about. The capture of these transcendent moments of our consciousness.⁸

到目前为止，围绕算法想象力的争论仍然牢牢扎根于此岸，而非柏林斯基假设和库兹韦尔早已预见的那个陌生空间。我们缺乏令人信服的证据表明算法具有意向性、创造力，或任何人们认为想象力所必需的特质。更大的问题是，我们无法真正跨越复杂的海洋，无法有效地理解那片隐喻性的浩瀚深渊，就像我们无法理解太平洋一样。我们具身化，根植于现实，受限于肉体和神经组织的生物物理实现，当我们在海洋中畅游时，我们只能体验到其中的一小部分。

So far, the debate around algorithmic imagination remains firmly grounded on this shore, not the alien space that Berlinski hypothesizes and Kurzweil already sees. We have no compelling evidence to suggest that algorithms have intentionality, creativity, or any of the traits one might consider necessary to an imaginative faculty. An even bigger problem is that we have no real way to voyage across the ocean of complexity, no way to effectively comprehend that metaphorical vasty deep any more than we can comprehend the Pacific. We are embodied, grounded, limited to biophysical implementations of flesh and nerve tissue, and when we swim in the ocean we can experience only a tiny part of it.

然而，通过同样的抽象和过程（在本例中，更类似于递归）的智力阶梯，我们可以自己想象算法想象力可能是什么样子。本书追溯了一系列与（可能的）富有想象力的算法的相遇，并注意到生物、文化和计算思维结构之间日益增长的认知交流。Google Now 和“预期”的设计目标都是一种富有想象力的思维形式——一个设想未来可能发生的事情，并通过实施将其转化为更大可能性的过程。谷歌认为我们可能会问的问题悬停在搜索栏的边缘，准备在我们开始提问后几毫秒内跳入视野。华尔街的算法也是如此，它们憧憬着市场未来并做出财务预测。

And yet, through the same intellectual ladders of abstraction and process (in this case, something more akin to recursion), we can imagine for ourselves what algorithmic imagination might be. This book has traced a series of encounters with (possible) imaginative algorithms, noting the growing cognitive traffic between biological, cultural, and computational structures of thinking. Google Now and the design goal of “anticipation” are forms of imaginative thinking—a process for envisioning possible futures and bringing them into greater possibility through implementation. The questions Google thinks we might ask hover at the edge of the search bar, ready to leap into view milliseconds after we begin to ask them. The algos of Wall Street do this as well, dreaming of market futures and making financial predictions.

我们只能通过人类可读的结果和输出来了解这些想象力的实践。但即使是分析看似最具创意的算法的输出，也无法真正揭示想象力在机器内部是如何运作的，尤其是在这些结果是专门为人类消费而定制的情况下。相反，我们必须运用这些投射和暗示的力量，来构想出一个认知工厂的景象，在那里这些输出被创造出来。这就是文化计算本身的空间，一个我们只能通过抽象、简化和采样才能进入的数学宇宙。我们无法完全理解驱动谷歌神经网络寻找类似眼睛的特征来增强的过程，也没有人类的经验与对同一幅图像进行数百万次迭代执行相同过程相关。推荐书籍、解读口语或执行其他数千项大数据分析任务的系统也是如此。除了最简单的情况外，我们永远无法知道算法是如何知道它们所知道的。这就是想象力的计算空间，或者在《萨曼莎夫人》中称之为“词语之间的空间”。

We know these practices of imagination only through their human-readable results and outputs. But analyzing the outputs of even the most seemingly creative algorithms does not really tell us how imagination works within the machine, particularly since the results are specifically tailored for human consumption. Instead, we must use those powers of projection and intimation to conjure up a vision of the cognitive factory where those outputs are created. This is the space of cultural computation itself, a mathematical universe that is accessible to us only through abstraction, simplification, and sampling. There is no way we can fully comprehend the processes driving Google’s neural networks as they hunt for eye-like features to enhance, no human experience that correlates to performing that same procedure on iterations of the same image millions of times. The same goes for the systems that recommend books, interpret spoken language, or perform any of a thousand other big data analytic tasks. Aside from the most simplistic cases, we will never know how algorithms know what they know. This is the computational space of imagination, or what in Her Samantha calls the “space between the words.”

远离海岸，算法想象力的深海不断将我们拉回自身，探寻驱动创造性工作的认知、灵感和机缘巧合的神秘起源。计算系统如何重塑、引导或调节这些过程？从个人层面来看，这直接延伸了技术：记忆库、虚拟助手或推荐引擎在创作过程中何时应得赞誉？这些工具为我们管理认知、灵感和机缘巧合，在我们的社交媒体流、数字工作空间和笔记本中，以及更广阔的可见知识领域中，创造对话和智力联系。作家使用文字处理器管理草稿；科学家使用研究数据库和引用工具管理专业知识领域；艺术家使用图像编辑软件、照片共享工具和虚拟笔记本来追踪观察——所有这些创造性过程都依赖于工具，这些工具在使用过程中日益活跃，有时甚至具有操纵性。如果算法阅读要求我们认识到我们的研究对象在我们解释它们时正在适应我们，那么算法想象就会问在什么时候这种适应过程会成为一种想象力的形式。

Going farther from shore, the deep waters of algorithmic imagination draw us relentlessly back toward ourselves and the mysterious origins of cognition, inspiration, and serendipity that drive creative work. How are computational systems reinventing, channeling, or modulating those processes? On an individual level this is a straightforward extension of technics: when does the memory bank, the virtual assistant, or the recommendation engine deserve credit in the creative process? These tools manage cognition, inspiration, and serendipity for us, generating conversation and intellectual connection in our social media streams, our digital workspaces and notebooks, and more broadly, in the horizon of visible knowledge. The writer using a word processor to manage drafts; the scientist using research databases and citation tools to manage a field of professional knowledge; the artist using image editing software, photo sharing tools, and a virtual notebook to track observations—all of these creative processes depend on tools that are increasingly active, occasionally manipulative agents in their own use. If algorithmic reading asks us to recognize that our objects of study are adapting to us as we interpret them, algorithmic imagination asks at what point that process of adaptation becomes a form of imaginative agency.

就像图灵提出的如何衡量智力一样，我们或许需要接受这样一个事实：想象力只能通过集体、主体间的方式进行衡量。技术如何赋能想象力的问题也延伸到集体想象的非算法过程：对话与协作网络，它使多个人能够同时取得同样的关键突破（例如微积分或声纳）。我们的算法机器现在不仅管理我们个人数据流，还管理着浩瀚的集体信息河流，它们像筛选我们的音乐或新闻一样定期筛选我们的朋友和同事。就匿名者、维基百科和无数其他集体创作模式而言，它们使全新的工作形式成为可能，将陌生人团结成集体，这些集体可以像通过 Facebook 组织的埃及抗议者一样具有自我意识，也可以像谷歌自动完成数据库的贡献者一样不自觉。这就是增强的想象力，是人类和机器只有共同才能完成的变革性工作。

Like Turing’s proposal for how to measure intelligence, we may need to accept that imagination can be measured only together, intersubjectively. This question of how technics enable imagination also extends to non-algorithmic processes of collective imagination: the networks of dialog and collaboration that allow multiple people to make the same critical breakthroughs in parallel (e.g., calculus or sonar). Our algorithmic machines now manage not just our individual streams of data but the great rivers of collective information, filtering our friends and colleagues as regularly as they curate our music or the news. In the case of Anonymous, Wikipedia, and myriad other modes of collective authorship, they make entire new forms of work possible, uniting strangers into collectives that can be as self-aware as Egyptian protestors organizing through Facebook or as unwitting as the contributors to Google’s autocomplete databases. This is the augmented imagination, the transformative work that humans and machines can only do together.

增强想象力

Augmenting Imagination

无论想象力是什么，我们都知道，我们工具的焦距会对其进行变形和改变。想想万尼瓦尔·布什（Vannevar Bush）对Memex的开创性设想吧。Memex是早期版本的星际迷航数字计算机，其灵感源于想象力增强的概念。诚然，它是一台通用的知识机器，但其大小和功能却与个人办公桌相当：“它是对……记忆的放大版私密补充。” ⁹布什首先触动了至今仍困扰着谷歌、苹果和算法先锋的共鸣：百科全书也必须是私密的。计算机必须使通用的东西不仅易于访问，而且个性化。Memex的主要功能是协助用户构建和审查“轨迹”，即文档之间的超文本关联，这些关联构成了一个个人的副文本层，就像一组地图，指引着你办公桌上那片微缩胶片的荒野。

Whatever imagination is, we know that the focal lenses of our tools inflect and change it. Consider Vannevar Bush’s seminal vision of the Memex, an early version of the digital Star Trek computer inspired by a conception of imaginative augmentation. It was a universal knowledge machine, to be sure, but with the size and function of a personal desk: “It is an enlarged intimate supplement to … memory.”⁹ Bush first struck the chord that continues to haunt Google, Apple, and the algorithmic vanguard today: the encyclopedia must also be intimate. Computers must make the universal not just accessible but personal. The chief function of the Memex would be to assist users in constructing and reviewing “trails,” or hypertextual associations between documents that constituted a personal paratextual layer, a set of maps for the wilderness of microfilm inhabiting one’s desk.

这条线索是布什对新兴信息时代核心挑战——选择问题的回应。他设想了一种新的职业——“开拓者”，这些人将在众多不同的文本之间建立最有用的路径和关联。知识工作将成为一种巧妙的选择和有针对性的省略的实践，使用户“有权忘记那些他不需要立即掌握的各种各样的事情，并保证如果它们变得重要，他可以再次找到它们”。但正如我们在本书中所看到的，这些开拓者越来越算法化，他们“在词语之间的空间中穿梭”，以便我们能够过上条理清晰的数字生活，不受原始数据海洋般嘈杂的干扰。如果说这些开拓者正在探索新的边疆，那么与之类比的并非狂野西部的侦察兵和定居者，而是如今正在穿越太阳系的自动探测器、机器人和探测车。这些使者与它们的人类创造者截然不同。

The trail was Bush’s response to the central challenge of the nascent information age, the question of selection. He imagines the new occupation of “trail blazer,” the people who would establish the most useful pathways and associate links across many different texts. Knowledge work would become the practice of artful selection and targeted omission, allowing the user “the privilege of forgetting the manifold things he does not need to have immediately at hand, with some assurance that he can find them again if they prove important.” But as we have seen throughout this book, the trail blazers are increasingly algorithmic, navigating “the space between the words” so we can live coherent digital lives untroubled by the oceanic cacophony of raw data. If these trail blazers are exploring a new frontier, the analogy is not to scouts and settlers of the Wild West but rather the automated probes, robots, and rovers now traversing the Solar System. These envoys are very different from their human creators.

布什认为小型化和自动化是Memex魔力的核心，其目的是为了支持一种非常人性化的数据检索和研究方式。布什设想中的Memex从未真正实现；个人存档和通用数据访问的功能如今实际上被分散在谷歌、维基百科、Dropbox和Evernote等平台和工具上。更重要的是，布什设想的作为成就巅峰的个人联想用例，即成为一名开拓者，如今却只是一个更深层次计算项目的副产品，该项目拥有其自身相关的算法想象空间，吸引了谷歌和其他信息领域巨头的顶尖工程师。那就是对人类知识的映射和操控，这是一项协作且过程性的巨大挑战，需要一套复杂的、相互关联的算法系统来处理从操作全球数据云到建模和预测人类行为的所有事务。

The miniaturization and automation that Bush saw as central to the magic of the Memex were intended to support a very human form of data retrieval and research. The Memex as Bush imagined it has never quite come into being; the functions of personal archive and universal data access are now effectively split across platforms and tools like Google, Wikipedia, Dropbox, and Evernote. More significantly, the personal, associative use case that Bush imagined as the pinnacle of achievement, being a trail blazer, has turned out to be a sideline product of a much deeper computational project with its own associated algorithmic space of imagination, one that engrosses top engineers at Google and other titans of the informational universe. That is the mapping and manipulating of human knowledge, a collaborative and processual grand challenge, which requires a sophisticated set of interlocking algorithmic systems to handle everything from operating the global data cloud to modeling and predicting human behavior.

Memex 的核心是一台想象机器，一种技术设备或“亲密补充”，旨在为联想思维、好奇心和创造力创造一个文字和智力空间。它试图设想数字信息访问在增强人类认知方面将发挥的作用。布什认为该系统的基础很平凡：百科全书、参考书以及作为每个用户个人数据领域基础的标准知识索引的汇编。他正确地预见到二战后的知识爆炸将使选择成为一项重要的操作，但他未能想象到这种系统性变革的速度。Memex 从未诞生，因为知识的基底不是可以机械操纵的静态场。它是一片不断运动的海洋。没有什么是安全的：百科全书不是（想想维基百科的兴起），科学教科书不是（必须每年更新），甚至人类知识的开创性著作也不是，它们不断地通过介入批评、发现和不断发展的社会背景进行补救。正如复杂性科学家 Sam Arbesman 在《事实的半衰期》一书中所说，绝大多数人类知识都是以这种方式偶然产生的，而认识论变革的速度正在加快。10^这主要是由于算法过程，它使研究人员能够提高工作效率，并创造出新的交叉授粉甚至自动化知识发现的方法。最重要的是，随着知识网络变得越来越复杂，我们越来越依赖算法对相关性的判断来帮助我们应对新事实和话语的指数级增长。对于布什来说，知识操作的界面就像 Memex 顶部的透明板一样薄，它将用户与存储其中的数千页缩微胶片隔开。对于我们来说，这个平面不断变厚，成为普适计算的“界面层”中的另一层膜。

The Memex was at its core an imagination machine, a technical apparatus or “intimate supplement” designed to create a literal and intellectual space for associative thinking, curiosity, and creativity. It attempted to envision the role that digital information access would have in augmenting human cognition. Bush saw the foundations of that system as unremarkable: encyclopedias, reference works, and the assemblage of a standard index of knowledge that would serve as the foundation for each user’s personal data sphere. He correctly apprehended that the explosion of knowledge after World War II would make selection a nontrivial operation, but he failed to imagine the rapidity of such systemic change. The Memex never came into being because the substrate of knowledge is not a static field to be manipulated mechanically. It is an ocean, constantly in motion. Nothing is safe: not the encyclopedia (consider the rise of Wikipedia), not the science textbooks (which must be updated yearly), not even the seminal works of human knowledge, which are constantly remediated by intervening criticism, discovery, and evolving social context. As complexity scientist Sam Arbesman argues in The Half-Life of Facts, the vast majority of human knowledge is contingent in this way, and the pace of epistemological change is accelerating.¹⁰ This is largely due to algorithmic processes, which enable researchers to be more productive and create new ways of cross-pollinating and even automating the discovery of knowledge. Most important, as the web of knowledge grows more complex, we become increasingly reliant on algorithmic judgments of relevance to help us navigate the exponential growth of new facts and discourses. For Bush, the interface for knowledge manipulation was as thin as the transparent plate on top of the Memex that separated users from the thousands of microfilm pages stored within it. For us, that plane continues to thicken, becoming another membrane in the “interface layer” of ubiquitous computation.

布什曾极具先见之明地设想了我们与数字信息之间将如何进行各种增强互动，但算法时代在普遍知识空间（Memex 内部）与在其表面构建知识轨迹的私密网络的人类用户之间引入了一层至关重要的抽象膜。谷歌的 PageRank 以及我们赖以管理信息海洋的无数其他系统的功能，在于不断地重新绘制，并最终重塑这片海洋。人类的角色并非策展人，而是重塑一个过时的数字隐喻：乘风破浪的冲浪者。我们身处一个复杂的空间，它拥有自身算法化的天气、洋流和能量来源，其中许多都被刻意地遮蔽了。如同冲浪者扫描海浪一样，我们与这些巨大力量的关系是情感的、本能的，有时甚至近乎原始的。这个比喻开辟了关于想象力的新视角：我们可以将增强人机想象力的可能性空间视为两个实体的交汇，这两个实体彼此理解非常有限，在截然不同的尺度上运作。冲浪者与海洋之间的动态张力正是这项活动的趣味所在。

Bush was deeply prescient in imagining the kinds of augmented interactions we would have with digital information, but the age of the algorithm has introduced this crucial membrane of abstraction between the space of universal knowledge (inside the Memex) and the human user constructing an intimate web of knowledge trails at its surface. The function of Google’s PageRank and countless other systems we depend on to curate the ocean of information is to continually remap, and ultimately to remake, that ocean. The role of the human is not one of curator but rather, to refashion an antiquated digital metaphor, a surfer riding the waves. We operate within a complex space that has its own algorithmic weather, currents, and sources of energy, many of them deliberately obscured from us. Like a surfer scanning the waves, our relationship with these vast forces is affective, visceral, and at times almost primordial. The metaphor opens up a view on imagination: we can think about the possibility space of augmented human–machine imagination as the intersection of two entities with very limited understanding of one another, operating at vastly different scales. That dynamic tension between the surfer and the ocean is just what makes the activity interesting.

两个愿望

The Two Desires

算法的故事始终充满着爱情。如同所有浪漫情缘，这种吸引力既源于熟悉，也源于陌生——既认同他人的自我，又对他人充满神秘感。作为技术系统，算法始终体现着我们自身的碎片：我们的记忆、我们的知识和信仰结构，以及我们的伦理和哲学基础。它们是人类意图和进步的镜子，映射出我们蕴含其中的显性和隐性知识。与此同时，它们也提供了神秘的本质要素，根据数据库的逻辑、复杂性和算法迭代进行操作，以人类根本无法理解的方式计算选择。

The story of the algorithm has always been a love affair. Like every romance, the attraction is based on both familiarity and foreignness—the recognition of ourselves in another as well as the mystery of that other. As technical systems, algorithms have always embodied fragments of ourselves: our memories, our structures of knowledge and belief, our ethical and philosophical foundations. They are mirrors for human intention and progress, reflecting back the explicit and tacit knowledge that we embed in them. At the same time they provide the essential ingredient of mystery, operating according to the logics of the database, complexity, and algorithmic iteration, calculating choices in ways that are fundamentally alien to human understanding.

表面上看，当代文化对算法的神化似乎并未让我们更接近实现我们对完美认知世界和完美认知自身的双重渴望。迄今为止，我们积累的万能百科全书浩如烟海，却又前后矛盾，充斥着矛盾、错误信息和空洞的引用。尽管神经科学、心理学、经济学和其他相关领域进行了数十年的研究，但在很多方面，我们对集体自我的理解似乎进展甚微。我们在网上数据库和用户资料中看到的自我形象仍然像是粗糙的漫画。算法神学所承诺的救赎依然遥不可及：计算层面在文化生活中笨拙而脱节的实现仍有许多不足之处。算法仍然经常出错，无法为超越真理提供令人信服的论据。

On the surface, it may seem as if the apotheosis of the algorithm in contemporary culture has not brought us any closer to the consummation of our twinned desires for perfect knowledge of the world and perfect knowledge of ourselves. The universal encyclopedia we have assembled to date is vast but inconsistent, full of contradictions, misinformation, and dangling references. In many ways our collective self-understanding seems to have advanced very little despite decades of research in neuroscience, psychology, economics, and other relevant fields. The versions of ourselves we see online in databases and user profiles still seem like crude caricatures. The promised salvation of algorithmic theology stubbornly remains in the distant future: the clunky, disjointed implementation of the computational layer on cultural life leaves much yet to be desired. Algorithms still get it wrong far too often to make a believable case for transcendent truth.

然而，诱惑依然存在。计算系统在掌握新实践领域（从理解自然语音到创作音乐）的每一步，都意味着人类在弥合差距方面迈出了一步。我们围绕代码的文化现实塑造自身，在计算的不足之处进行补强，并积极地扩展它，最终完成无处不在的算法大厦。其中一些选择很简单，甚至有些迂腐，比如调整我们的说话方式（我们希望）以便机器更容易理解我们的语句。另一些选择则更加微妙，比如为了拍出最佳自拍机会而安排周末的诱惑，或者名义上客观的代码中隐藏的偏见。我们越来越依赖计算系统来获取知识生活的原始素材，从书籍和新闻报道到最基本的词汇、分享的想法以及分享的对象。我们对这些文化机器投入越多，就越能在合作的道路上走得更远。这不仅仅是合作：而是一种共同认同。我们开始通过数字化实践来定义我们是谁，因为虚拟空间对我们来说比内在空间更加真实。

And yet the seduction remains. For every step that computational systems take in mastering a new field of practice, from understanding natural speech to composing music, humanity also takes a step to close the gap. We shape ourselves around the cultural reality of code, shoring up the facade of computation where it falls short and working feverishly to extend it, to complete the edifice of the ubiquitous algorithm. Some of these choices are simple, even pedantic, like adjusting our speech patterns to (we hope) make our statements easier for machines to understand. Others are far more subtle, like the temptation to organize one’s weekend for optimal selfie opportunities, or the hidden biases encoded in nominally objective code. We depend on computational systems for a growing share of the raw material of intellectual life, from books and news stories to the very basics, like vocabulary, ideas to share, and people to share them with. The more we invest ourselves in these culture machines, the further we proceed down a path of collaboration. More than collaboration: a kind of co-identity. We are coming to define who we are through digital practice because virtual spaces are becoming more real to us than visceral ones.

以谷歌及其对星际迷航计算机的追求为例。该公司在查找和索引全球信息方面取得了巨大进步。我们许多人依赖谷歌，不仅是为了搜索和访问，还将其作为各种个人数据的存储库，从电子邮件和照片，到我们运动和饮食习惯的生物特征测量。该公司在存储和部署这些不同形式信息方面的成功，不仅使其自身的系统更接近星际迷航的目标，也使其用户更接近目标。研究人员已经证明，使用谷歌会改变我们的记忆习惯，任何超过30岁的人都可以证明这一点：只需问问自己，与拥有第一部手机之前相比，你现在记住了多少个电话号码。11^使用文字处理器进行写作、社交媒体对身份形成的影响等等也是如此。12^这些都是对巨变的细微观察，随着我们阅读、写作、对话和思考的核心实践成为数字化过程，数字文化、记忆和身份正在以多种方式演变。通过大脑的可塑性和不断变化的社会规范，我们正在使自己变得更加容易被算法机器所了解。

Take the example of Google and the quest for the Star Trek computer. The company has made tremendous strides in finding and indexing the world’s information. Many of us rely on Google not just for search and access but as a repository for all sorts of personal data, from emails and photographs to biometric measurements of our exercise and diet routines. The company’s success at storing and deploying these different forms of information does more than just move its own systems closer to the Star Trek goal—it also moves its users closer. Researchers have demonstrated that using Google changes our memory practices, a fact anyone over thirty can prove: just ask yourself how many phone numbers you remember now as opposed to before you owned your first cell phone.¹¹ The same is true for composition using word processors, for the impact of social media on identity formation, and so on.¹² These are small observations of the sea change, the many ways in which digital culture, memory, and identity are evolving as our core practices of reading, writing, conversation, and thinking become digital processes. Through brain plasticity and changing social norms, we are adapting ourselves to become more knowable for algorithmic machines.

正如海尔斯等人所言，我们正与技术系统共同进化，缓慢地走向算法之恋的某种圆满结局。灾难的风险萦绕在我们心头——这种圆满结局可能演变成一场碰撞，一场爆炸，就像我们在《终结者》系列等故事以及未来生命研究所（由埃隆·马斯克等人资助，旨在避免《终结者》带来的人工智能末日）等机构中反复提及的那样。但也有更乐观的愿景，即人类与计算系统进行富有成效的合作：《星际迷航》式的未来，或者像科幻作家伊恩·班克斯的《文化》系列小说那样，一个更加雄心勃勃、由人工智能驱动的社会。事实上，我们花费了太多时间担忧叛逆的独立人工智能的崛起，以至于我们很少停下来思考，我们已经在以多种方式与各种智能的自主系统进行合作。这远远超出了我们对数字地址簿、邮件程序或文件档案的依赖：谷歌的机器学习算法现在可以对电子邮件提出适当的回复，而 AlphaGo 为这项古老艺术形式的大师们提供了一些最有趣的游戏。

In this way we are evolving, as Hayles and others have argued, in conjunction with our technical systems, slowly moving toward some consummation of the algorithmic love affair. The risk of disaster haunts us—the consummation might become a collision, an explosion of the kinds we linger on in stories like the Terminator series and through institutions like the Future of Life Institute (funded by Elon Musk and others to avert the Terminator AI apocalypse). But there is a more optimistic vision as well, one where humans engage in productive collaboration with computational systems: the Star Trek future, or a more ambitiously AI-fueled society like science fiction author Iain Banks’s Culture novels. Indeed, we spend so much time worrying about the rise of a renegade independent artificial intelligence that we rarely pause to consider the many ways in which we are already collaborating with autonomous systems of varied intelligence. This moves far beyond our reliance on digital address books, mail programs, or file archives: Google’s machine learning algorithms can now suggest appropriate responses to emails, and AlphaGo gives grandmasters of that venerable art form some of their most interesting games.

进一步拓展视野，我们开始看到人类正在如何改变认知和想象的基本术语。算法时代标志着技术记忆进化到不仅存储数据，还存储更为复杂的实践模式，从音乐品味到社交图谱。在很多情况下，我们已经在与机器协同想象。算法系统策划着对知识的探索，与我们的兴趣和信息需求进行对话和预测。它们与我们共同创作，为从《纸牌屋》到经过算法审核的流行音乐等各种事物提供框架、语境，有时甚至提供直接素材。想象力的可能性范围日益由计算系统决定，它们制造并策划着推动思想、话语和艺术生命周期的偶然发现和信息流。换句话说，主宰文明史的双重探索，正日益与蕴含在有效可计算性中的渴望共同决定。想象的空间存在于算法环境中，无法计算的东西就无法完全融入我们现在生活的更广泛的文化结构中。

Widening the scope further, we can begin to see how we are changing the fundamental terms of cognition and imagination. The age of the algorithm marks the moment when technical memory has evolved to store not just our data but far more sophisticated patterns of practice, from musical taste to our social graphs. In many cases we are already imagining in concert with our machines. Algorithmic systems curate the quest for knowledge, conversing and anticipating our interests and informational needs. They author with us, providing scaffolding, context, and occasionally direct material for everything from House of Cards to algorithmically vetted pop music. The horizon of imaginative possibility is increasingly determined by computational systems, which manufacture and curate the serendipity and informational flow that propels the lifecycle of ideas, of discourse, of art. In other words, the twin quests that have dominated the history of civilization are increasingly codeterminate with the desire embedded in effective computability. The space of imagination exists in algorithmic context, and that which cannot be computed cannot be fully integrated into the broader fabric of culture as we live it now.

我们正在努力应对代码带来的后果，因为许多人类经验和文化工作的边界案例困扰着当代算法文化。策展人、编辑和评论家的角色比以往任何时候都更加重要，因为我们正在划定有效可计算性的界限，并努力学习和记住我们的计算系统所不知道或无法知道的事情。

We are grappling with the consequences of code through the many boundary cases of human experience and cultural work that trouble contemporary algorithmic culture. The role of the curator, the editor, and the critic is more important than ever, as we draw the lines of effective computability and struggle to learn and remember the things that our computational systems do not or cannot know.

实验人文学科

Experimental Humanities

我们可以选择将算法的形象视为值得崇拜的神明……或者，我们可以选择在我们的文化游戏中看到一个新的参与者、合作者和对话者。这就是我所说的“实验人文学科”，它源于我们在本书中提出的同一哲学起点。理论与实践、抽象与现实之间的差距，是人类参与的关键点。随着我们的社会钟摆越来越偏向计算主义阵营，这为人文学者提供了一个绝佳的切入点。

We can choose to construe the figure of the algorithm as a god to be worshipped … or we can choose to see a new player, collaborator, and interlocutor in our cultural games. This is what I would like to call “experimental humanities,” and it springs from the same philosophical starting point we began with in this book. The gap between theory and implementation, between abstraction and reality, is a critical point of human engagement. It represents a tremendous opportunity for humanists to take the field as our societal pendulum swings ever farther into the camp of computationalism.

我想以提出一种公共人文研究的愿景来结束本书，这种研究将直接与算法文化互动。通过打破批评与创造力、观察与行动、阅读与写作之间虚幻的壁垒，我们可以定义一个基于“文化机器”概念的深度参与文化生活的新模式。作为一种学术探究模式，实验人文学科将批判性嵌入到研究过程中，利用算法机器（包括计算型和其他类型的机器）与生俱来的协作潜力，使新的集体辩论、洞察和理解形式成为可能。这一过程的第一步是超越我们作为现有系统的消费者，或者充其量是用户的角色。如果我们让文化机器设定批判框架，我们最终将完全生活在“有效可计算”的空间中，忘记边界标记之外的一切。我们需要一种实验人文学科，以实践、理论和教学法的有效结合来回应当前学术界对人文学科的批判：构建我们自己的文化机器。

I want to close this book by advancing a vision of public, humanistic research that engages directly with algorithmic culture. By breaking down the illusory wall between criticism and creativity, between observation and action, between reading and writing, we can define a new model for deep engagement with cultural life founded on the notion of the culture machine. As a mode of scholarly inquiry, experimental humanities embeds critique in process, using the innate collaborative potential of algorithmic machines (both computational and otherwise) to make new forms of collective debate, insight, and understanding possible. The first step in this process is moving beyond our roles as consumers or, at best, users of existing systems. If we let our culture machines set the critical frame, we will wind up living entirely in the space of the “effectively computable,” forgetting everything beyond the boundary markers. We need an experimental humanities that answers current critiques of humanities in the academy with an energetic combination of praxis, theory, and pedagogy: building culture machines of our own.

从务实的角度来说，认识到这种与快速变化的知识世界之间的关系，对于重新构建人文学科的专业实践至关重要。经典、文学领域，甚至个别书籍，都不再是曾经稳定的知识实体，因为参与保存和研究它们的机构也经历了同样快速的技术变革，而这些变革也影响着算法海洋的其他部分。探索和管理计算空间是一项极富想象力的项目。

In pragmatic terms, recognizing this relationship with a rapidly changing universe of knowledge is vital to reframing the professional practice of the humanities. Canons, literary fields, even individual books are no longer the stable intellectual entities they once were, as the institutions involved in their preservation and study undergo the same rapid technological changes affecting the rest of the algorithmic ocean. The work of navigating and curating computational space is a deeply imaginative project.

当我们这样做，与算法文本、平台甚至对话者协同扮演批判性角色时，我们也在进行着增强想象。利用计算框架来扩展人类认知空间的新型批判性工作有着巨大的潜力。这听起来或许有些乌托邦，但此类实践已成为数字文化体验中司空见惯的一部分。我们利用评论和标签讲述集体笑话和故事，构建共享的叙事和杂糅，这些叙事和杂糅本身就能演变成复杂的技术实体，成为像#lolcats这样肤浅的网络表情包，或者像#blacklivesmatter这样有影响力的网络表情包。这些都是集体增强的时刻，利用数字平台以前所未有的方式吸引注意力和共识。这种新兴的集体、即兴想象渠道已经开始将较为沉稳的文化实践从24小时新闻周期转变为学术会议。

When we do this, performing critical roles in concert with algorithmic texts, platforms, and even interlocutors, we are also performing augmented imagination. There is great potential for new forms of critical work that use computational scaffolding to extend the space of human cognition. This may sound utopian, but such practices have become a quotidian part of digital cultural experience. We tell collective jokes and stories using comment threads and hashtags, building shared narratives and farragoes that can evolve into sophisticated technical beings in their own right as Internet memes as superficial as #lolcats or as potent as #blacklivesmatter. These are moments of collective augmentation, leveraging digital platforms to build attention and consensus in ways that were previously impossible. This emerging channel for collective, ad hoc imagination has started to transform more staid cultural practices from the twenty-four-hour news cycle to the scholarly conference.

举几个例子，我们可以从算法系统正在忙于颠覆的记忆与遗忘实践入手。医学和科学领域的研究人员现在可以合理地预期，自动化系统能够自动交叉引用并将新研究置于更广泛的关键领域中，提供共引索引和量化的广泛相关性指标。人文学科的学者也即将迎来类似的时刻，届时我们可以可靠地依赖算法系统来记住我们已经忘记或从未知晓的已发表著作的学术联系。这些系统可以扩展，以增强对语境的理解和与文本资源的深度互动，提供指向被引文献来源的档案链接（使读者能够直接评估我们如今很容易在脚注中得出的推论）。几年后，我们可以想象一下，将学术话语的不同层次的“审查”整合到一份数字重写本中，将多种模式的学术副文本带入或并入“最终”发表的研究成果中，从会议或社交媒体流中的即兴评论到同行评审本身的反馈，这也许并不荒谬。

To offer a few examples, we can begin with the practices of memory and forgetting that algorithmic systems are busily upending. Researchers in medicine and the sciences can now reasonably expect automated systems to cross-reference and contextualize new research within broader critical fields automatically, providing co-citation indexes and quantitatively broad measures of relevance. Scholars in the humanities are on the cusp of a similar moment, when we can reliably depend on algorithmic systems to remember intellectual connections to published work that we have forgotten, or never knew. These same systems can be extended to enhance contextual understanding and deep engagement with textual resources, providing archival links to the sources of cited documents (allowing the reader to directly evaluate the inferences we so easily slide into footnotes today). In a few years it may not be absurd to imagine different layers of “vetting” for scholarly discourse being layered into a single digital palimpsest, bringing multiple modes of scholarly paratext into or alongside the “final” published research, from off-the-cuff remarks at a conference or social media stream to the feedback of peer review itself.

批判代码研究工作组的参与者采用熟悉的细读和深入的语境审查策略，但在计算抽象层面的较低层级，他们将软件本身视为文化文本。这一框架可以产生关于抽象与实现之间差距的惊人见解，例如通过揭示模拟城市犯罪率计算的意识形态假设，正如数字文化学者马克·桑普尔所做的那样。13^这些批评家通过在协作平台上进行讨论以及通过电子书评论和数字人文季刊等开放获取平台发表，以算法方式进行辩论。事实上，在ebr上编辑批判代码研究讨论的一大挑战（我曾在那里担任过一段时间的主题编辑）是捕捉计算介导的讨论论坛交流的活力并将其提炼为更传统的固定文本格式。

The participants in the Critical Code Studies Working Group adopt familiar tactics of close reading and intensive contextual scrutiny, but at a lower rung in the layer of computational abstraction, reading software itself as cultural text. This frame can generate breathtaking insights into the gap between abstraction and implementation, for example by revealing the ideological assumptions of SimCity’s crime rate calculations, as digital culture scholar Mark Sample has done.¹³ These critics perform their debates algorithmically, through discussion on collaborative platforms and publication via open-access platforms like the electronic book review and Digital Humanities Quarterly. Indeed, one of the great challenges of editing the Critical Code Studies discussions for publication on ebr, where I served for a time as a thread editor, was capturing the vibrancy of computationally mediated discussion forum exchanges and distilling it into a more traditionally fixed textual format.

其他实验人文主义者（如果我可以给他们起这个称号的话）则通过相反的方法实现了同样的目标，他们用创作代替阅读，从而为了抗议和批评的目的而重新发明了算法系统。亚利桑那州立大学的“活力生活”项目利用现场表演和装置艺术来探索数据生产的内在后果以及“我们如何评价人、艺术品和信息”。14^像Sample 的 @NSA_prismbot 这样的批判性推特机器人的创造者正在推出他们自己的算法，在集体话语空间中提出伦理问题。15^这些机器人和“活力生活”都将企业数据流转化为数字媒体学者 Rita Raley 所说的“战术媒体”的空间，以强大的即时性参与社会批评。这就是算法批评：这些项目涉及代码、表演和复杂的观众参与形式，从而构建了集体对话。

Other experimental humanists, if I can offer them the title, achieve the same goal by taking the opposite approach, authoring instead of reading, and thereby reinventing algorithmic systems for the purposes of protest and critique. The Vibrant Lives project based at Arizona State University uses live performance and installations to explore the visceral consequences of data production and “how we value persons, art objects, and information.”¹⁴ The creators of critical Twitter bots like Sample’s @NSA_prismbot are launching algorithms of their own that pose ethical questions in spaces of collective discourse.¹⁵ These bots and Vibrant Lives both transform corporate data streams into spaces for what digital media scholar Rita Raley calls “tactical media,” engaging in social critique with a powerful immediacy. This is algorithmic criticism: the projects involve code, performance, and sophisticated forms of audience engagement that build collective conversation.

桑普尔基于推特的项目也引发了人们对批判诚信问题的关注。桑普尔的机器人总是已被攻陷并牵连，它们使用着与他所模仿的实体相同的企业宣传工具。但这却是我们所有人的现状，我们在自身知识事业的每个阶段都深深地被算法系统所裹挟。本书的写作过程必然产生了成千上万次谷歌搜索，如果没有数百种有用的算法和计算平台，这一切都不可能实现。我们都在算法面前进行批判。大海已然拍打着我们的脚跟；我们只需转身，直面它。

Sample’s Twitter-based project also calls attention to the question of critical integrity. Sample’s bots are always already compromised and implicated, using the same corporate publicity tools as the entities he is parodying. But this is the state of play for all of us, who are deeply enmeshed in algorithmic systems at every stage of our own intellectual enterprises. The process of writing this book must have generated thousands of Google queries and would have been impossible without access to hundreds of helpful algorithms and computational platforms. We are all performing criticism in the presence of the algorithm. The ocean is already lapping at our heels; we need only to turn and face it ourselves.

布什设想的Memex实际上是一个盒子里的图书馆，鼓励用户标记文本。信息产业正在开发实现这些功能的工具，但通常都隐藏在围墙花园之后，比如亚马逊的Kindle注释功能，众所周知，很难从其专有的本地格式中提取出来。我们依赖于那些由碎片化信息产业空间的杂乱商业逻辑驱动的工具——它们仍然足够诱人，足够接近布什的愿景，以至于诱使我们忘记我们错过了什么。想象一下，如果有一个强大的跨学科实验系统，我们可以做什么：例如，一个用于集体注释和讨论文学文本的当代平台，或者一个在高强度会议或研讨会后立即生成内容丰富的互联研究文档的系统。

The Memex that Bush envisioned was effectively a library in a box, one that encouraged its users to mark up the texts. The information industry is developing tools that serve these functions, but often behind walled gardens, like Amazon’s Kindle annotations, which are famously difficult to extract from their proprietary home format. We are dependent on tools that are driven by the hodgepodge commercial logic of a fragmented industrial space for information—they remain alluring enough, close enough to Bush’s vision to tempt us into forgetting what we are missing. Imagine what we might do with robust systems for interdisciplinary, experimental work: a contemporary platform for the collective annotation and discussion of literary texts, for example, or a system for generating richly interlinked research documents immediately after a high-energy conference or workshop.

越来越多的重要学术工作不再局限于图书馆，而是通过社交媒体和面对面交流进行。捕捉此类学术活动的能量和细微之处，并有效地提炼、整理和保存其鲜活之处，供信息宇宙中的其他旅行者阅读，将变得更加重要。

An increasing share of the most important scholarly work takes place not in solitary libraries but through social media and in-person exchanges. It will only become more vital to capture the energy and nuance of such performances of scholarship in ways that effectively distill, curate, and preserve their liveliness for other travelers in the universe of information.

我认为，接受我们与知识的关系已成为一套应对日益复杂的互动实践，也意味着承认人文学科在算法时代的重要作用。模糊性、不和谐、诠释、情感：这些都是人文和艺术探究的舞台。当我们通过算法门户访问知识世界时，它本身就是一种主动的媒介，塑造着我们凝视的框架和基调，读者的主体地位就变得愈发重要。长期以来，人文学科一直在努力理解本书开篇所讲述的故事：语言的神秘力量，文字和代码的魔力。我们迫切需要更多的读者，更多的批评家，来解读那些如今定义着我们集体想象的渠道和视野的算法。

To accept that our relationship with knowledge has become a set of practices for interacting with rapidly increasing complexity is, I would argue, also an acceptance of the vital role of the humanities in the age of the algorithm. Ambiguity, dissonance, interpretation, affect: these are the playing fields of humanistic and artistic inquiry. The subject position of the reader becomes increasingly important when the algorithmic portal through which we are accessing the universe of knowledge is itself an active agent, shaping the framing and the tenor of our gaze. The humanities has long grappled with the story that started this book: the mythic power of language, the incantatory magic of words and codes. We desperately need more readers, more critics, to interpret the algorithms that now define the channels and horizons of our collective imaginations.

笔记

Notes

• 堕胎，64
• Abortion, 64
• 抽象，10
  • 美学与，83，87–112
  • 套利和 161
  • Bogost 和，49，92–95
  • 资本主义和，165
  • 背景和 24
  • 加密货币和 160–180
  • 培养机器和，54（另见培养机器）
  • 控制论和 28, 30, 34
  • 渴望得到答案，25
  • 丢弃的信息，以及 50
  • 有效可计算性，28，33
  • 信息精神，159
  • 高频交易（HFT）和
  • 想象力和，185，189，192，194
  • 接口和，52，54，92，96，103，108，110-111
  • 阶梯，82–83
  • 语言和，2，24
  • 马克思主义与，165
  • 意义和，36
  • 金钱和，153，159，161，165-167，171-175
  • Netflix 和，87–112，205n36
  • 政治，45
  • 实用主义方法，19–21
  • 过程和，2，52，54
  • 现实与，205n36
  • Siri 和 64–65、82–84
  • 图灵机和，23（另见图灵机）
  • Uber 和，124–126，129
  • 维纳和，28–29，30
  • 算法的工作，113，120，123–136，139–149
• Abstraction, 10
  • aesthetics and, 83, 87–112
  • arbitrage and, 161
  • Bogost and, 49, 92–95
  • capitalism and, 165
  • context and, 24
  • cryptocurrency and, 160–180
  • culture machines and, 54 (see also Culture machines)
  • cybernetics and, 28, 30, 34
  • desire for answer and, 25
  • discarded information and, 50
  • effective computability and, 28, 33
  • ethos of information and, 159
  • high frequency trading (HFT) and
  • imagination and, 185, 189, 192, 194
  • interfaces and, 52, 54, 92, 96, 103, 108, 110–111
  • ladder of, 82–83
  • language and, 2, 24
  • Marxism and, 165
  • meaning and, 36
  • money and, 153, 159, 161, 165–167, 171–175
  • Netflix and, 87–112, 205n36
  • politics of, 45
  • pragmatist approach and, 19–21
  • process and, 2, 52, 54
  • reality and, 205n36
  • Siri and, 64–65, 82–84
  • Turing Machine and, 23 (see also Turing Machine)
  • Uber and, 124–126, 129
  • Wiener and, 28–29, 30
  • work of algorithms and, 113, 120, 123–136, 139–149
• 亚当斯，道格拉斯，123
• Adams, Douglas, 123
• 亚当斯，亨利，80–81
• Adams, Henry, 80–81
• 自适应系统，50，63，72，92，174，176，186，191
• Adaptive systems, 50, 63, 72, 92, 174, 176, 186, 191
• 成瘾，114–115，118–119，121–122，176
• Addiction, 114–115, 118–119, 121–122, 176
• AdSense，158–159
• AdSense, 158–159
• 算法的出现（柏林斯基），9，24
• Advent of the Algorithm, The (Berlinski), 9, 24
• 广告
  • AdSense 和 158–159
  • 算法套利，111，161
  • 苹果和，65
  • 等待的文化计算，34
  • 作为文化潜伏期，159
  • 情感诉求，148
  • Facebook 和 113–114
  • 反馈系统，145–148
  • 谷歌和，66，74，156，158-160
  • 哈贝马斯，175
  • Netflix 和 98、100、102、104、107–110
  • Uber 和 125
• Advertisements
  • AdSense and, 158–159
  • algorithmic arbitrage and, 111, 161
  • Apple and, 65
  • cultural calculus of waiting and, 34
  • as cultural latency, 159
  • emotional appeals of, 148
  • Facebook and, 113–114
  • feedback systems and, 145–148
  • Google and, 66, 74, 156, 158–160
  • Habermas on, 175
  • Netflix and, 98, 100, 102, 104, 107–110
  • Uber and, 125
• 美学
  • 抽象与，83，87–112
  • 套利和，109–112，175
  • 文化机器和55
  • 《纸牌屋》和，92，98–112
  • Netflix量子理论与，91–97
  • 个性化和，11，97–103
  • 生产，12
  • 算法的工作，123，129，131，138–147
• Aesthetics
  • abstraction and, 83, 87–112
  • arbitrage and, 109–112, 175
  • culture machines and, 55
  • House of Cards and, 92, 98–112
  • Netflix Quantum Theory and, 91–97
  • personalization and, 11, 97–103
  • of production, 12
  • work of algorithms and, 123, 129, 131, 138–147
• 阿格雷，菲利普，178–179
• Agre, Philip, 178–179
• Airbnb，124，127
• Airbnb, 124, 127
• 代数，17
• Algebra, 17
• 算法阅读，52-56
• Algorithmic reading, 52–56
• 算法交易，12，20，99，155
• Algorithmic trading, 12, 20, 99, 155
• 算法
  • 抽象和，2（另见抽象）
  • 套利和，12，51，97，110-112，119，121，124，127，130-134，140，151，160，162，169，171，176
  • 柏林斯基路、9、24、30、36、181
  • 比特币和，160–180
  • 黑匣子和，7、15–16、47–48、51、55、64、72、92–93、96、136、138、146–147、153、162、169–171、179
  • 区块链和 163–168、171、177、179
  • Bogost 和，16，33，49
  • 丘奇-图灵论题及23–26、39–41、73
  • 意识和, 2, 4, 8, 22–23, 36–37, 40, 76–79, 154, 176, 178, 182, 184
  • DARPA 和，11，57–58，87
  • 欲望和，21–26, 37, 41, 47, 49, 52, 79–82, 93–96, 121, 159, 189–192
  • 有效可计算性，10、13、21–29、33–37、40–49、52–54、58、62、64、72–76、81、93、192–193
  • 椭圆曲线数字签名算法和，163
  • 体现和，26–32
  • 加密，153，162-163
  • 框架和，118–119
  • 启蒙运动与，27，30，38，45，68-71，73
  • 实验人文学科，192–196
  • Facebook 和 20（另见Facebook）
  • 信仰和, 7–9, 12, 16, 78, 80, 152, 162, 166, 168
  • 游戏化，12、114–116、120、123–127、133
  • 机器里的幽灵，55，95
  • 停止状态，41–46
  • 高频交易（HFT）以及151–158、168–169、177
  • 如何思考，36–41
  • 意识形态和，7、9、18、20–23、26、33、38、42、46–47、54、64、69、130、144、155、160–162、167、169、194
  • 想象力和，11，55–56，181–196
  • 实施情况，47-52
  • 智能助理，11、57、62、64–65、77
  • 亲密关系，4、11、35、54、65、74–78、82–85、97、102、107、128–130、172、176、185–189
  • Knuth 和，17–18
  • 语言和，24–28，33–41，44，51，54–55
  • 机器学习和 2、15、28、42、62、66、71、85、90、112、181–184、191
  • 数理逻辑和2
  • 意义和，35–36，38，44–45，50，54–55
  • 隐喻和，32–36
  • Netflix 奖及，87–91
  • 神经网络和，28，31，39，182–183，185
  • 单向函数，162–163
  • 实用主义方法，18–25，42，58，62
  • 过程和，41–46
  • 可编程培养和，169–175
  • 追求完美的知识，13，65，71，73，190
  • 文化机器的兴起，15–21（另见文化机器）
  • Siri 和 59（另请参阅Siri）
  • 旅行商问题和
  • 图灵机和，9（另见图灵机）
  • 作为计算的载体，5
  • 想要，81–85
  • Weizenbaum 和，33–40
  • 作品，113–149
  • 崇拜，192
• Algorithms
  • abstraction and, 2 (see also Abstraction)
  • arbitrage and, 12, 51, 97, 110–112, 119, 121, 124, 127, 130–134, 140, 151, 160, 162, 169, 171, 176
  • Berlinski on, 9, 24, 30, 36, 181
  • Bitcoin and, 160–180
  • black boxes and, 7, 15–16, 47–48, 51, 55, 64, 72, 92–93, 96, 136, 138, 146–147, 153, 162, 169–171, 179
  • blockchains and, 163–168, 171, 177, 179
  • Bogost and, 16, 33, 49
  • Church-Turing thesis and, 23–26, 39–41, 73
  • consciousness and, 2, 4, 8, 22–23, 36–37, 40, 76–79, 154, 176, 178, 182, 184
  • DARPA and, 11, 57–58, 87
  • desire and, 21–26, 37, 41, 47, 49, 52, 79–82, 93–96, 121, 159, 189–192
  • effective computability and, 10, 13, 21–29, 33–37, 40–49, 52–54, 58, 62, 64, 72–76, 81, 93, 192–193
  • Elliptic Curve Digital Signature Algorithm and, 163
  • embodiment and, 26–32
  • encryption, 153, 162–163
  • enframing and, 118–119
  • Enlightenment and, 27, 30, 38, 45, 68–71, 73
  • experimental humanities and, 192–196
  • Facebook and, 20 (see also Facebook)
  • faith and, 7–9, 12, 16, 78, 80, 152, 162, 166, 168
  • gamification and, 12, 114–116, 120, 123–127, 133
  • ghost in the machine and, 55, 95
  • halting states and, 41–46
  • high frequency trading (HFT) and, 151–158, 168–169, 177
  • how to think about, 36–41
  • ideology and, 7, 9, 18, 20–23, 26, 33, 38, 42, 46–47, 54, 64, 69, 130, 144, 155, 160–162, 167, 169, 194
  • imagination and, 11, 55–56, 181–196
  • implementation and, 47–52
  • intelligent assistants and, 11, 57, 62, 64–65, 77
  • intimacy and, 4, 11, 35, 54, 65, 74–78, 82–85, 97, 102, 107, 128–130, 172, 176, 185–189
  • Knuth and, 17–18
  • language and, 24–28, 33–41, 44, 51, 54–55
  • machine learning and, 2, 15, 28, 42, 62, 66, 71, 85, 90, 112, 181–184, 191
  • mathematical logic and, 2
  • meaning and, 35–36, 38, 44–45, 50, 54–55
  • metaphor and, 32–36
  • Netflix Prize and, 87–91
  • neural networks and, 28, 31, 39, 182–183, 185
  • one-way functions and, 162–163
  • pragmatist approach and, 18–25, 42, 58, 62
  • process and, 41–46
  • programmable culture and, 169–175
  • quest for perfect knowledge and, 13, 65, 71, 73, 190
  • rise of culture machines and, 15–21 (see also Culture machines)
  • Siri and, 59 (see also Siri)
  • traveling salesman problem and
  • Turing Machine and, 9 (see also Turing Machine)
  • as vehicle of computation, 5
  • wants of, 81–85
  • Weizenbaum and, 33–40
  • work of, 113–149
  • worship of, 192
• 花剌子米，阿布·阿卜杜拉·穆罕默德·本·穆萨，17 岁
• Al-Khwārizmī, Abū ‘Abdullāh Muhammad ibn Mūsā, 17
• Alphabet公司，66，155
• Alphabet Corporation, 66, 155
• AlphaGo，182，191
• AlphaGo, 182, 191
• 亚马逊
  • 算法套利，124
  • 人工智能（AI）和135–145
  • 贝佐斯和174
  • 比特币和169
  • 商业模式，20–21，93–94
  • 云仓库和 131–132、135–145
  • 颠覆性技术，以及 124
  • 有效可计算性，以及 42
  • 效率算法和134
  • 界面经济和124
  • Kindle 和 195
  • Kiva Systems 和 134
  • Mechanical Turk 和 135–145
  • 个性化和 97
  • 实物物流，13，131
  • 采摘者和，132–134
  • 务实的方法，18
  • 产品改进，以及 42
  • 机器人技术，134
  • 简化精神，97
  • 工人条件和，132–134，139–140
• Amazon
  • algorithmic arbitrage and, 124
  • artificial intelligence (AI) and, 135–145
  • Bezos and, 174
  • Bitcoin and, 169
  • business model of, 20–21, 93–94
  • cloud warehouses and, 131–132, 135–145
  • disruptive technologies and, 124
  • effective computability and, 42
  • efficiency algorithms and, 134
  • interface economy and, 124
  • Kindle and, 195
  • Kiva Systems and, 134
  • Mechanical Turk and, 135–145
  • personalization and, 97
  • physical logistics of, 13, 131
  • pickers and, 132–134
  • pragmatic approach and, 18
  • product improvement and, 42
  • robotics and, 134
  • simplification ethos and, 97
  • worker conditions and, 132–134, 139–140
• 安卓，59
• Android, 59
• 匿名，112，186
• Anonymous, 112, 186
• 美国在线（AOL），75岁
• AOL, 75
• 苹果，81岁
  • 增强想象力，186
  • 黑匣子，169
  • 云仓库，131
  • 公司价值，158
  • 有效可计算性，以及 42
  • 效率算法和134
  • 富士康和，133–134
  • 全球计算基础设施，131
  • iOS App Store 和 59{tab}
  • iTunes 和 161
  • 大规模基础设施，131
  • 本体论和，62–63，65
  • 实物物流，131
  • 实用主义方法，18
  • 产品改进，以及 42
  • 可编程文化，169
  • 搜索和, 87
  • Siri 和 57（另见Siri）
  • 软件和，59，62
  • SRI国际和，57，59
• Apple, 81
  • augmenting imagination and, 186
  • black box of, 169
  • cloud warehouse of, 131
  • company value of, 158
  • effective computability and, 42
  • efficiency algorithms and, 134
  • Foxconn and, 133–134
  • global computation infrastructure of, 131
  • iOS App Store and, 59{tab}
  • iTunes and, 161
  • massive infrastructure of, 131
  • ontology and, 62–63, 65
  • physical logistics of, 131
  • pragmatist approach and, 18
  • product improvement and, 42
  • programmable culture and, 169
  • search and, 87
  • Siri and, 57 (see also Siri)
  • software and, 59, 62
  • SRI International and, 57, 59
• 应用程序接口 (API)，7，113
• Application Program Interfaces (APIs), 7, 113
• 应用程序
  • 文化机器和15
  • Facebook和，9，113–115，149
  • 她和，83
  • 身份和 6
  • 接口和，8，124，145
  • iOS App Store 和 59
  • Lyft 和 128, 145
  • Netflix 和 91、94、102
  • 第三方，114–115
  • Uber 和 124, 145
• Apps
  • culture machines and, 15
  • Facebook and, 9, 113–115, 149
  • Her and, 83
  • identity and, 6
  • interfaces and, 8, 124, 145
  • iOS App Store and, 59
  • Lyft and, 128, 145
  • Netflix and, 91, 94, 102
  • third-party, 114–115
  • Uber and, 124, 145
• 阿拉伯之春，111，186
• Arab Spring, 111, 186
• 阿贝斯曼，塞缪尔，188–189
• Arbesman, Samuel, 188–189
• 套利
  • 算法，12，51，97，110–112，119，121，124，127，130–134，140，151，160，162，169，171，176
  • 比特币和，51，169–171，175–179
  • 文化, 12, 94, 121, 134, 152, 159
  • 不同的价值观，121–122
  • Facebook 和 111
  • 谷歌和111
  • 高频交易（HFT）以及151–158、168–169、177
  • 界面经济和，123–131，139–140，145，147
  • 劳动和，97，112，123–145
  • 市场问题，以及 152、161
  • 采矿价值和，176–177
  • 金钱和，151–152，155–163，169–171，175–179
  • Netflix 和 94、97、109–112
  • PageRank 和 159
  • 定价，12
  • 实时，12
  • 胜过内容，13
  • 重视文化，155–160
• Arbitrage
  • algorithmic, 12, 51, 97, 110–112, 119, 121, 124, 127, 130–134, 140, 151, 160, 162, 169, 171, 176
  • Bitcoin and, 51, 169–171, 175–179
  • cultural, 12, 94, 121, 134, 152, 159
  • differing values and, 121–122
  • Facebook and, 111
  • Google and, 111
  • high frequency trading (HFT) and, 151–158, 168–169, 177
  • interface economy and, 123–131, 139–140, 145, 147
  • labor and, 97, 112, 123–145
  • market issues and, 152, 161
  • mining value and, 176–177
  • money and, 151–152, 155–163, 169–171, 175–179
  • Netflix and, 94, 97, 109–112
  • PageRank and, 159
  • pricing, 12
  • real-time, 12
  • trumping content and, 13
  • valuing culture and, 155–160
• 阿基米德，18岁
• Archimedes, 18
• 人工智能（AI）
  • 自适应系统和，50，63，72，92，174，176，186，191
  • 亚马逊和，135–145
  • 拟人化和，83，181
  • 预期和，73–74
  • 人造的，135–141
  • 自动机和，135–138
  • DARPA 和，11，57–58，87
  • 深蓝和，135–138
  • DeepMind 和 28、66、181–182
  • 欲望和，79–82
  • ELIZA 和，34
  • 机器里的幽灵，55，95
  • HAL和181
  • 稳态和，199n42
  • 人类大脑，29
  • 思想史，61
  • 智能助理，11、57、62、64–65、77
  • 亲密关系，75–76
  • 裁员，以及 133
  • McCulloch-Pitts神经元和，28，39
  • 机器学习和 2、15、28、42、62、66、71、85、90、112、181–186
  • Mechanical Turk 和，12，135–145
  • 自然语言处理（NLP）以及 62–63
  • 神经网络和，28，31，39，182–183，185
  • OS One（她）和，77
  • 叛逆的独立人士，191
  • 萨曼莎（她）和，77–85，154，181
  • Siri 和，57，61（另见Siri）
  • 图灵测试和，43，79-82，87，138，142，182
• Artificial intelligence (AI)
  • adaptive systems and, 50, 63, 72, 92, 174, 176, 186, 191
  • Amazon and, 135–145
  • anthropomorphism and, 83, 181
  • anticipation and, 73–74
  • artificial, 135–141
  • automata and, 135–138
  • DARPA and, 11, 57–58, 87
  • Deep Blue and, 135–138
  • DeepMind and, 28, 66, 181–182
  • desire and, 79–82
  • ELIZA and, 34
  • ghost in the machine and, 55, 95
  • HAL and, 181
  • homeostat and, 199n42
  • human brain and, 29
  • intellectual history of, 61
  • intelligent assistants and, 11, 57, 62, 64–65, 77
  • intimacy and, 75–76
  • job elimination and, 133
  • McCulloch-Pitts Neuron and, 28, 39
  • machine learning and, 2, 15, 28, 42, 62, 66, 71, 85, 90, 112, 181–186
  • Mechanical Turk and, 12, 135–145
  • natural language processing (NLP) and, 62–63
  • neural networks and, 28, 31, 39, 182–183, 185
  • OS One (Her) and, 77
  • renegade independent, 191
  • Samantha (Her) and, 77–85, 154, 181
  • Siri and, 57, 61 (see also Siri)
  • Turing test and, 43, 79–82, 87, 138, 142, 182
• 《计算机编程艺术》（Knuth），17
• Art of Computer Programming, The (Knuth), 17
• 阿什比，罗斯，199n42
• Ashby, Ross, 199n42
• 阿西莫夫，艾萨克，45岁
• Asimov, Isaac, 45
• 大西洋杂志，7，92，170
• Atlantic, The (magazine), 7, 92, 170
• 自动化，122，134，144，188
• Automation, 122, 134, 144, 188
• 自创生，28–30
• Autopoiesis, 28–30

• 查尔斯·巴贝奇，8岁
• Babbage, Charles, 8
• 伊恩·班克斯，191
• Banks, Iain, 191
• 巴尼特，贝琳达，43–44
• Barnet, Belinda, 43–44
• 贝叶斯分析，182
• Bayesian analysis, 182
• BBC，170
• BBC, 170
• BellKor 的《务实混沌》（Netflix），89–90
• BellKor’s Pragmatic Chaos (Netflix), 89–90
• 大卫·柏林斯基, 9, 24, 30, 36, 181, 184
• Berlinski, David, 9, 24, 30, 36, 181, 184
• 杰夫·贝佐斯，174岁
• Bezos, Jeff, 174
• 大数据，11，15–16，62–63，90，110
• Big data, 11, 15–16, 62–63, 90, 110
• 生物学，2，4，26–33，36–37，80，133，139，185
• Biology, 2, 4, 26–33, 36–37, 80, 133, 139, 185
• 比特币，12-13
  • 套利和，51，169–171，175–179
  • 区块链和 163–168、171–172、177、179
  • 计算主义方法和
  • 文化加工和 178
  • 消除脆弱性，161–162
  • 椭圆曲线数字签名算法和，163
  • 加密和，162–163
  • 玻璃盒，162
  • 内在价值，165
  • 劳动和，164，178
  • 合法性，以及 178
  • 市场问题，163–180
  • 矿工和，164–168，171–172，175–179
  • 中本聪等，161–162，165–167
  • 单向函数，162–163
  • 可编程培养和，169–175
  • 交易费用，以及164–165
  • 透明度和，160–164，168，171，177–178
  • 信任和，166–168
• Bitcoin, 12–13
  • arbitrage and, 51, 169–171, 175–179
  • blockchains and, 163–168, 171–172, 177, 179
  • computationalist approach and
  • cultural processing and, 178
  • eliminating vulnerability and, 161–162
  • Elliptic Curve Digital Signature Algorithm and, 163
  • encryption and, 162–163
  • as glass box, 162
  • intrinsic value and, 165
  • labor and, 164, 178
  • legitimacy and, 178
  • market issues and, 163–180
  • miners and, 164–168, 171–172, 175–179
  • Nakamoto and, 161–162, 165–167
  • one-way functions and, 162–163
  • programmable culture and, 169–175
  • transaction fees and, 164–165
  • transparency and, 160–164, 168, 171, 177–178
  • trust and, 166–168
• 百视达，99
• Blockbuster, 99
• 区块链，163–168，171–172，177，179
• Blockchains, 163–168, 171–172, 177, 179
• 博客
  • 早期网络策划，156
  • Facebook 算法和 178
  • Gawker Media 和，170–175
  • 新闻原则和，173，175
  • 采矿价值和，175，178
  • Netflix 和 91–92
  • 特克的工作条件和，139
  • Uber 和 130
• Blogs
  • early web curation and, 156
  • Facebook algorithms and, 178
  • Gawker Media and, 170–175
  • journalistic principles and, 173, 175
  • mining value and, 175, 178
  • Netflix and, 91–92
  • turker job conditions and, 139
  • Uber and, 130
• 布鲁姆，哈罗德，175
• Bloom, Harold, 175
• 伊恩·博格斯特
  • 抽象和，92–95
  • 算法和，16，33，49
  • 计算大教堂，6–8、27、33、49、51
  • 计算和，6–10，16
  • Cow Clicker和，12，116–123
  • 启蒙运动和，8
  • 游戏化，12、114–116、120、123–127、133
  • Netflix 和 92–95
• Bogost, Ian
  • abstraction and, 92–95
  • algorithms and, 16, 33, 49
  • cathedral of computation and, 6–8, 27, 33, 49, 51
  • computation and, 6–10, 16
  • Cow Clicker and, 12, 116–123
  • Enlightenment and, 8
  • gamification and, 12, 114–116, 120, 123–127, 133
  • Netflix and, 92–95
• 布尔与非，51
• Boolean conjunctions, 51
• 比安卡·博斯克（Bianca Bosker），58岁
• Bosker, Bianca, 58
• 尼克·博斯特罗姆，45岁
• Bostrom, Nick, 45
• 鲍克，杰弗里，28岁，110
• Bowker, Geoffrey, 28, 110
• 博克斯利修道院，1​​37
• Boxley Abbey, 137
• 《大脑采摘》（波波娃），175
• Brain Pickings (Popova), 175
• 大脑可塑性，38，191
• Brain plasticity, 38, 191
• 布兰德，斯图尔特，3，29
• Brand, Stewart, 3, 29
• 巴西（电影），142
• Brazil (film), 142
• 绝命毒师（电视剧），101
• Breaking Bad (TV series), 101
• 布林，谢尔盖，57，155–156
• Brin, Sergei, 57, 155–156
• 巴菲特，沃伦，174
• Buffett, Warren, 174
• 伯尔，雷蒙德，95岁
• Burr, Raymond, 95
• 布什，万尼瓦尔，18，186–189，195
• Bush, Vannevar, 18, 186–189, 195
• 商业模式
  • 亚马逊和，20–21，93–94，96
  • 加密货币和 160–180
  • Facebook 和 20
  • FarmVillage和 115
  • 谷歌和，20–21，71–72，93–94，96，155，159
  • Netflix 和 87–88
  • Uber 和 54、93–94、96
• Business models
  • Amazon and, 20–21, 93–94, 96
  • cryptocurrency and, 160–180
  • Facebook and, 20
  • FarmVille and, 115
  • Google and, 20–21, 71–72, 93–94, 96, 155, 159
  • Netflix and, 87–88
  • Uber and, 54, 93–94, 96
• 《启蒙事业》（达恩顿）
  • 68，68
• Business of Enlightenment, The (Darnton)
  • 68, 68

• 微积分，24，26，30，34，44–45，98，148，186
• Calculus, 24, 26, 30, 34, 44–45, 98, 148, 186
• 卡罗, 57–58, 63, 65, 67, 79, 81
• CALO, 57–58, 63, 65, 67, 79, 81
• 坎贝尔，约瑟夫，94岁
• Campbell, Joseph, 94
• 坎贝尔·默里，138
• Campbell, Murray, 138
• 《资本主义》，12，105
  • 加密货币和，160，165–168，170–175
  • 假装，146–147
  • Gawker Media 和，170–175
  • 身份和，146–147
  • 界面经济和，127，133
  • 劳动和，165
  • 公共领域，172–173
  • 风险投资，9，124，174
• Capitalism, 12, 105
  • cryptocurrency and, 160, 165–168, 170–175
  • faking it and, 146–147
  • Gawker Media and, 170–175
  • identity and, 146–147
  • interface economy and, 127, 133
  • labor and, 165
  • public sphere and, 172–173
  • venture, 9, 124, 174
• 捕捉术，113
• Captology, 113
• 卡尔·尼古拉斯 38岁
• Carr, Nicholas, 38
• 卡鲁斯，艾莉森，131
• Carruth, Allison, 131
• 卡斯特罗诺瓦，爱德华，121
• Castronova, Edward, 121
• 大教堂和集市（雷蒙德），6
• Cathedral and the Bazaar, The (Raymond), 6
• 计算大教堂，6–10、27、33、49、51
• Cathedral of computation, 6–10, 27, 33, 49, 51
• 国际象棋，135–138，144–145
• Chess, 135–138, 144–145
• Chun, Wendy Hui Kyong, 3, 16, 33, 35–36, 42, 104
• Chun, Wendy Hui Kyong, 3, 16, 33, 35–36, 42, 104
• 丘奇，阿隆佐，23–24，42
• Church, Alonzo, 23– 24, 42
• 丘奇-图灵论题，23–26，39–41
• Church-Turing thesis, 23–26, 39–41
• Cinematch (Netflix)，88–90，95
• Cinematch (Netflix), 88–90, 95
• 公民联合会案，174
• Citizens United case, 174
• 克拉克，安迪，37，39–40
• Clark, Andy, 37, 39–40
• 云仓库
  • 亚马逊和，135–145
  • 界面经济和，131–145
  • Mechanical Turk 和 135–145
  • 工人条件和，132–134，139–140
• Cloud warehouses
  • Amazon and, 135–145
  • interface economy and, 131–145
  • Mechanical Turk and, 135–145
  • worker conditions and, 132–134, 139–140
• 美国有线电视新闻网，170
• CNN, 170
• 代码。另请参阅算法
  • 计算大教堂，6–8、27、33、49、51
  • 计算主义方法，21，24–25，43，46，52
  • 作为语言，4，44（另见语言）
  • 魔法，1–5, 8, 10, 16, 49–50, 196
  • 意义和，36
• Code. See also Algorithms
  • cathedral of computation and, 6–8, 27, 33, 49, 51
  • computationalist approach and, 21, 24–25, 43, 46, 52
  • as language, 4, 44 (see also Language)
  • as magic, 1–5, 8, 10, 16, 49–50, 196
  • meaning and, 36
• 认识
  • CALO 和 57–58、63、65、67、79、81
  • 丘奇-图灵论题和39
  • 体现空间，49
  • 增强型，55
  • 延伸思维，37，40，43
  • 人类, 5, 25, 36, 39, 41, 55, 154
  • 想象力和，182，185，188–193
  • 修改，42
  • 新模式，51
  • 现实和，5
  • 认可和 36
  • 寻找意义，36
  • Siri 和 57–65、71–84
  • 技术, 26, 39, 154
  • 工具，40
• Cognition
  • CALO and, 57–58, 63, 65, 67, 79, 81
  • Church-Turing thesis and, 39
  • embodied space and, 49
  • enhanced, 55
  • extended mind and, 37, 40, 43
  • human, 5, 25, 36, 39, 41, 55, 154
  • imagination and, 182, 185, 188–193
  • modification of, 42
  • new patterns in, 51
  • reality and, 5
  • recognition and, 36
  • search for meaning and, 36
  • Siri and, 57–65, 71–84
  • technical, 26, 39, 154
  • tools of, 40
• 柯勒律治，塞缪尔·泰勒，209n20
• Coleridge, Samuel Taylor, 209n20
• 计算
  • 抽象和，19（另见抽象）
  • 自适应系统和，50，63，72，92，174，176，186，191
  • 应用程序和, 6, 8–9, 15, 59, 83, 91, 94, 102, 113–114, 124, 128, 145, 149
  • 大数据和，11，15–16，62–63，90，110，185
  • 博格斯特，6–10，16，27，33，49，51
  • 大教堂，6–10，27，33，49，51
  • 丘奇-图灵论题及23–26、39–41、73
  • 认知和，36，51（另见认知）
  • 意识和, 2, 4, 8, 22–23, 36–37, 40, 76–79, 154, 176, 178, 182, 184
  • 文化和，6，13，46，51–52，59，122，135，175
  • 欲望和，21–26, 37, 41, 47, 49, 52, 79–82, 93–96, 121, 159, 189–192
  • 有效可计算性，10、13、21–29、33–37、40–49、52–54、58、62、64、72–76、81、93、192–193
  • 机器里的幽灵，55，95
  • 停止状态，41–46
  • 实施情况，47-52
  • 智能助理，11、57、62、64–65、77
  • 接口和，8，96，110（另见接口）
  • 亲密关系，4、11、35、54、65、74–78、82–85、97、102、107、128–130、172、176、185–189
  • 内在价值，165
  • 逻辑与，2（另见逻辑）
  • 机器学习和 2、15、28、42、62、66、71、85、90、112、181–184、191
  • 记忆和，18，21，37，43–44，51，56，58，69，75，159–160，176，185–186，191–193
  • 摩尔定律和 43
  • 神话和，26（另见神话）
  • 本体论和8
  • 政治，53
  • 实用主义方法，18–25，42，58，62
  • 追求完美的知识，13，65，71，73，190
  • 偶然出现的故障，以及 55
  • 社会价值，12
  • 空间，2–5，9，21，42，45，76，154，185
  • 结构，41，48，122，185
  • 系统，3、5、8、10、15–17、20、22、25、27、30、33、44、49–51、55–58、93、104、110、122、142、151、176、185、190–192
  • 有形的，4
  • 理论，21，52
  • 文化转型，2
  • 图灵和，6–9、23–30、33、39–43、54、73、79–82、87、138、142、182
  • 无处不在，3–4，15，33，43，54，119，124–125，127，178，189–190
  • Wiener 和，16、26–32、34、37、39、43、57、122
• Computation
  • abstraction and, 19 (see also Abstraction)
  • adaptive systems and, 50, 63, 72, 92, 174, 176, 186, 191
  • apps and, 6, 8–9, 15, 59, 83, 91, 94, 102, 113–114, 124, 128, 145, 149
  • big data and, 11, 15–16, 62–63, 90, 110, 185
  • Bogost on, 6–10, 16, 27, 33, 49, 51
  • cathedral of, 6–10, 27, 33, 49, 51
  • Church-Turing thesis and, 23–26, 39–41, 73
  • cognition and, 36, 51 (see also Cognition)
  • consciousness and, 2, 4, 8, 22–23, 36–37, 40, 76–79, 154, 176, 178, 182, 184
  • culture and, 6, 13, 46, 51–52, 59, 122, 135, 175
  • desire and, 21–26, 37, 41, 47, 49, 52, 79–82, 93–96, 121, 159, 189–192
  • effective computability and, 10, 13, 21–29, 33–37, 40–49, 52–54, 58, 62, 64, 72–76, 81, 93, 192–193
  • ghost in the machine and, 55, 95
  • halting states and, 41–46
  • implementation and, 47–52
  • intelligent assistants and, 11, 57, 62, 64–65, 77
  • interfaces and, 8, 96, 110 (see also Interfaces)
  • intimacy and, 4, 11, 35, 54, 65, 74–78, 82–85, 97, 102, 107, 128–130, 172, 176, 185–189
  • intrinsic value and, 165
  • logic and, 2 (see also Logic)
  • machine learning and, 2, 15, 28, 42, 62, 66, 71, 85, 90, 112, 181–184, 191
  • memory and, 18, 21, 37, 43–44, 51, 56, 58, 69, 75, 159–160, 176, 185–186, 191–193
  • Moore’s Law and, 43
  • myth and, 26 (see also Myth)
  • ontology and, 8
  • politics of, 53
  • pragmatist approach and, 18–25, 42, 58, 62
  • quest for perfect knowledge and, 13, 65, 71, 73, 190
  • serendipitous glitches and, 55
  • social value and, 12
  • space of, 2–5, 9, 21, 42, 45, 76, 154, 185
  • structures for, 41, 48, 122, 185
  • systems of, 3, 5, 8, 10, 15–17, 20, 22, 25, 27, 30, 33, 44, 49–51, 55–58, 93, 104, 110, 122, 142, 151, 176, 185, 190–192
  • tangible, 4
  • theory of, 21, 52
  • transformation of culture and, 2
  • Turing and, 6–9, 23–30, 33, 39–43, 54, 73, 79–82, 87, 138, 142, 182
  • ubiquitous, 3–4, 15, 33, 43, 54, 119, 124–125, 127, 178, 189–190
  • Wiener and, 16, 26–32, 34, 37, 39, 43, 57, 122
• 计算主义方法
  • 比特币和171
  • 有效可计算性，21、24–25、43、46、52
  • 终局，45
  • 哥伦比亚和，38
  • 硬索赔，22
  • 意识形态，42
  • 想象力和，183–184，192
  • 信息标准，28
  • 机器学习和 183
  • Netflix 和 90、104
  • 本体论和，8，22
  • 政治批评，46
  • 可编程性和 16
  • 硅谷和，49
  • Siri 和 65、77
  • 图灵机和，23
  • 愿景，43
• Computationalist approach
  • Bitcoin and, 171
  • effective computability and, 21, 24–25, 43, 46, 52
  • endgame of, 45
  • Golumbia and, 38
  • hard claim for, 22
  • ideology of, 42
  • imagination and, 183–184, 192
  • information yardstick for, 28
  • machine learning and, 183
  • Netflix and, 90, 104
  • ontology and, 8, 22
  • political critique of, 46
  • programmability and, 16
  • Silicon Valley and, 49
  • Siri and, 65, 77
  • Turing Machine and, 23
  • vision of, 43
• 计算神权政治，7，9
• Computational theocracy, 7, 9
• 意识
  • 连贯的主观性，76–77
  • 计算和，2，4，8，22–23，36–37，40，76–79，154，176，178，182，184
  • 对话和 78
  • 延伸思维，37，40，43
  • 人类，8，37
  • 库兹韦尔，184
  • 建模，23
  • 叙事结构和 154
  • 本体论和178
  • 萨曼莎（她）和，77–85，154
  • 第二级，78
  • Siri 和 57–65、71–84
  • 溪流，176
  • 图灵机和，23，184
  • 图灵测试和，43，79-82，87，138，142，182
  • 上传，36
• Consciousness
  • coherent subjectivity and, 76–77
  • computation and, 2, 4, 8, 22–23, 36–37, 40, 76–79, 154, 176, 178, 182, 184
  • conversation and, 78
  • extended mind and, 37, 40, 43
  • human, 8, 37
  • Kurzweil on, 184
  • modeling, 23
  • narrative construction and, 154
  • ontology and, 178
  • Samantha (Her) and, 77–85, 154
  • second level of, 78
  • Siri and, 57–65, 71–84
  • stream of, 176
  • Turing Machine and, 23, 184
  • Turing test and, 43, 79–82, 87, 138, 142, 182
  • uploading, 36
• 融合文化（詹金斯），102
• Convergence Culture (Jenkins), 102
• 对话，85
  • 建筑集体，195
  • 意识和，78
  • 数据挖掘和 175
  • 作为数字化过程，191
  • 政策，132
  • Siri 和 57–65、71–84
  • 星际迷航电脑和，67
  • 图灵机和，182（另见图灵机）
  • Uber 和 130
• Conversation, 85
  • building collective, 195
  • consciousness and, 78
  • data-mining and, 175
  • as digital process, 191
  • policies for, 132
  • Siri and, 57–65, 71–84
  • Star Trek computer and, 67
  • Turing Machine and, 182 (see also Turing Machine)
  • Uber and, 130
• 康威，约翰，29–30
• Conway, John, 29–30
• 每日精彩网站，156，171
• Cool site of the day, 156, 171
• 奶牛点击器（游戏）
  • 荒谬，116，118
  • Bogost 和，12，116–123
  • 文化机器和，119–120
  • 故意设计不良，118–119
  • 框架和，118–119
  • 游戏化，12、114–116、120、123–127、133
  • 乐趣的反转，120–123
  • 意义和，116，118–119
  • 作为对社交游戏的回应，116
  • 规则，116
• Cow Clicker (game)
  • absurdity of, 116, 118
  • Bogost and, 12, 116–123
  • culture machines and, 119–120
  • deliberate poor design of, 118–119
  • framing and, 118–119
  • gamification and, 12, 114–116, 120, 123–127, 133
  • inversion of fun and, 120–123
  • meaning and, 116, 118–119
  • as response to social games, 116
  • rules of, 116
• 关键代码研究工作组, 194
• Critical Code Studies Working Group, 194
• 计算的文化逻辑（Golumbia），45
• Cultural Logic of Computation, The (Golumbia), 45
• 文化小说（班克斯），191
• Culture novels (Banks), 191
• 文化机器
  • 抽象层和 124–126
  • 美学和，55
  • 博格斯特和，7
  • CALO 和 57–58、63、65、67、79、81
  • 核心职能，53
  • 合作和，13，190
  • 计算，51，122，135，175
  • 上下文和 11
  • 奶牛响片和，119–120
  • 作为策展人，112
  • 数据云和 131
  • 体现和，10，26-32，80，82
  • 经验，34–35
  • Facebook 和，111（另见Facebook）
  • 基于信仰，7，175
  • 游戏化，115–116
  • Google 和 74（另见Google）
  • 高频交易（HFT）以及151–158、168–169、177
  • 人类物质性，49
  • 人类作为延伸，135
  • 意识形态和，47
  • 想象力和，55，147
  • 亲密关系，129
  • 劳动和，93，119
  • 语言和，39–40
  • LCARS（《星际迷航》）和，67–68
  • 机器学习和 2、15、28、42、62、66、71、85、90、112、181–184
  • 数学和，49–50
  • Mechanical Turk 和，12，135–145
  • 媒体学者和 54
  • 采矿价值和，175–176
  • 电影租赁和，88–89（另见Netflix）
  • 本体论和，62–65，68，69
  • 多孔性质，48
  • 处理和 42
  • 与文化的关系，26，47，48，193
  • 崛起，15–21
  • 萨曼莎（她）和，77–85，154
  • Siri 和 58（另请参阅Siri）
  • 社会行动和，52，148
  • 术语的使用，2
  • 价值挖掘和 116
• Culture machines
  • abstraction layers and, 124–126
  • aesthetics and, 55
  • Bogost and, 7
  • CALO and, 57–58, 63, 65, 67, 79, 81
  • central functions of, 53
  • collaboration and, 13, 190
  • computational, 51, 122, 135, 175
  • context and, 11
  • Cow Clicker and, 119–120
  • as curators, 112
  • data cloud and, 131
  • embodiment and, 10, 26–32, 80, 82
  • experience and, 34–35
  • Facebook and, 111 (see also Facebook)
  • faith-based, 7, 175
  • gamification and, 115–116
  • Google and, 74 (see also Google)
  • high frequency trading (HFT) and, 151–158, 168–169, 177
  • human materiality and, 49
  • humans as extensions of, 135
  • ideology and, 47
  • imagination and, 55, 147
  • intimacy and, 129
  • labor and, 93, 119
  • language and, 39–40
  • LCARS (Star Trek) and, 67–68
  • machine learning and, 2, 15, 28, 42, 62, 66, 71, 85, 90, 112, 181–184
  • mathematics and, 49–50
  • Mechanical Turk and, 12, 135–145
  • media scholars and, 54
  • mining value and, 175–176
  • movie rentals and, 88–89 (see also Netflix)
  • ontology and, 62–65, 68, 69
  • porous nature of, 48
  • processing and, 42
  • relationship with culture and, 26, 47, 48, 193
  • rise of, 15–21
  • Samantha (Her) and, 77–85, 154
  • Siri and, 58 (see also Siri)
  • social action and, 52, 148
  • use of term, 2
  • value mining and, 116
• 控制论
  • 抽象和，28，30，34
  • 自创生，28–30
  • 意识，41
  • 后果，122–123
  • 有效可计算性，28–29，34，37
  • 体现和，10，26-32，80，82
  • McCulloch-Pitts神经元和，28，39
  • 摩尼教体系和，122–123
  • 作为沟通的隐喻，32
  • 飞蛾机和，31–32，34
  • 第二波，28–29
  • 符号逻辑和 2, 21, 26
  • 从确定性到概率的转变，27、30、37
  • 理解世界，33
  • Wiener 和，15、26–32、34、37、39、42、57、122
• Cybernetics
  • abstraction and, 28, 30, 34
  • autopoiesis and, 28–30
  • consciousness and, 41
  • consequences of, 122–123
  • effective computability and, 28–29, 34, 37
  • embodiment and, 10, 26–32, 80, 82
  • McCulloch-Pitts Neuron and, 28, 39
  • Manichean systems and, 122–123
  • as metaphor for communication, 32
  • moth machine and, 31–32, 34
  • second wave of, 28–29
  • symbolic logic and, 2, 21, 26
  • transition from certainty to probability and, 27, 30, 37
  • understanding world and, 33
  • Wiener and, 15, 26–32, 34, 37, 39, 42, 57, 122
• 控制论（维纳），26–27
• Cybernetics (Wiener), 26–27
• 赛博朋克，3
• Cyperpunk, 3

• 让·达朗贝尔, 68–69, 74, 157
• D’Alembert, Jean, 68–69, 74, 157
• 达恩顿，罗伯特，68–69
• Darnton, Robert, 68–69
• 数据挖掘
  • 比特币和，165，167–172
  • Gawker Media 和 172
  • 意义和，175
  • 评分系统和 21
  • 价值和，175–179
• Data-mining
  • Bitcoin and, 165, 167–172
  • Gawker Media and, 172
  • meaning and, 175
  • scoring systems and, 21
  • value and, 175–179
• 决策，20，28，34，37，90
• Decision-making, 20, 28, 34, 37, 90
• 深蓝，135–138
• Deep Blue, 135–138
• DeepMind，28，66，181–182
• DeepMind, 28, 66, 181–182
• 国防高级研究计划局（DARPA），11，57-58，87
• Defense Advanced Research Projects Agency (DARPA), 11, 57–58, 87
• 德勒兹，吉尔斯，76岁
• Deleuze, Gilles, 76
• 登顿，尼克，170–174
• Denton, Nick, 170–174
• 笛卡尔，勒内，26，69，75
• Descartes, René, 26, 69, 75
• 丹斯切尔·祖伊（59岁）
• Deschanel, Zooey, 59
• 欲望
  • 人工智能（AI）和，79–82
  • 计算和，21–26, 37, 41, 47, 49, 52, 79–82, 93–96, 121, 159, 189–192
  • 消费者，93–96，121，159
  • 有效可计算性，13，52
  • 想象力和，189–192
  • Netflix 和 93–96
  • 图灵机和，25
• Desire
  • artificial intelligence (AI) and, 79–82
  • computation and, 21–26, 37, 41, 47, 49, 52, 79–82, 93–96, 121, 159, 189–192
  • consumer, 93–96, 121, 159
  • effective computability and, 13, 52
  • imagination and, 189–192
  • Netflix and, 93–96
  • Turing Machine and, 25
• 丹尼斯·狄德罗，11, 45, 57, 68–69, 72, 74, 156–157
• Diderot, Denis, 11, 45, 57, 68–69, 72, 74, 156–157
• 数字文化，3、7、18、22、43、49、66、87、156、160、191、193–194
• Digital culture, 3, 7, 18, 22, 43, 49, 66, 87, 156, 160, 191, 193–194
• 数字人文季刊（期刊），194
• Digital Humanities Quarterly (journal), 194
• 《方法论》（笛卡尔），69
• Discourse on the Method (Descartes), 69
• 歧视，21，130
• Discrimination, 21, 130
• 多明戈斯，佩德罗，183
• Domingos, Pedro, 183
• Dropbox，188
• Dropbox, 188
• 戴森，乔治，6岁
• Dyson, George, 6

• 复活节彩蛋，60，148
• Easter eggs, 60, 148
• eBay，121
• eBay, 121
• 艾柯，翁贝托，157
• Eco, Umberto, 157
• 有效可计算性
  • 抽象和，28，33
  • 算法和，10，13，21–29，33–37，40–49，52–54，58，62，64，72–76，81，93，192–193
  • 自动化科学和 44
  • 丘奇-图灵论题及23–24
  • 计算主义方法，21，24–25，43，46，52
  • 控制论和，28–29，34，37
  • 欲望和，13，52
  • 体现和，26–29
  • 可替代空间，35
  • 如何思考，36–41
  • 想象力和，46，93，192–193
  • 实施，47–49，54
  • 数学和，40
  • 隐喻和34
  • Netflix 和 93
  • 实用主义方法，25–26
  • 流程方法论，以及 53
  • 证明和，24
  • 寻求普遍知识，13
  • 科学实验，以及 44
  • Siri 和 58、62、64、72–76、81
  • 转型概念，21
  • 图灵机和，41–42
  • 魏森鲍姆，37–38
• Effective computability
  • abstraction and, 28, 33
  • algorithms and, 10, 13, 21–29, 33–37, 40–49, 52–54, 58, 62, 64, 72–76, 81, 93, 192–193
  • automated science and, 44
  • Church-Turing thesis and, 23–24
  • computationalist approach and, 21, 24–25, 43, 46, 52
  • cybernetics and, 28–29, 34, 37
  • desire and, 13, 52
  • embodiment and, 26–29
  • fungible space of, 35
  • how to think of, 36–41
  • imagination and, 46, 93, 192–193
  • implementation and, 47–49, 54
  • mathematics and, 40
  • metaphor and, 34
  • Netflix and, 93
  • pragmatist approach and, 25–26
  • process methodology and, 53
  • proofs and, 24
  • quest for universal knowledge and, 13
  • scientific experimentation and, 44
  • Siri and, 58, 62, 64, 72–76, 81
  • transformational concept of, 21
  • Turing Machine and, 41–42
  • Weizenbaum on, 37–38
• 爱因斯坦，阿尔伯特，181
• Einstein, Albert, 181
• 伊丽莎，34岁
• ELIZA, 34
• 椭圆曲线数字签名算法，163
• Elliptic Curve Digital Signature Algorithm, 163
• 体现，10，26-32，80，82
• Embodiment, 10, 26–32, 80, 82
• 加密，153，162-163
• Encryption, 153, 162–163
• 百科全书, 45, 68–73, 84, 157, 172
• Encyclopédie, 45, 68–73, 84, 157, 172
• 《我们所知的探索的终结》（辛格哈尔），72
• “End of Search as We Know It, The” (Singhal), 72
• 框架，118–119
• Enframing, 118–119
• 启示
  • 算法和，27，30，38，45，68–71，73
  • 博格斯特和，8
  • 第四等级，174
  • 想象力和，184
  • 新的客观性，175
  • 追求完美的知识，13，65，71，73，190
  • Siri 和，7–8、11、71–76、79–80、82
  • 维纳和，30
• Enlightenment
  • algorithms and, 27, 30, 38, 45, 68–71, 73
  • Bogost and, 8
  • fourth estate and, 174
  • imagination and, 184
  • new objectivity and, 175
  • quest for perfect knowledge and, 13, 65, 71, 73, 190
  • Siri and, 7–8, 11, 71–76, 79–80, 82
  • Wiener and, 30
• 内森·恩斯门格（Nathan Ensmenger），17岁
• Ensmenger, Nathan, 17
• 企业家，6, 18, 123, 127, 135, 148, 161, 169, 175
• Entrepreneurs, 6, 18, 123, 127, 135, 148, 161, 169, 175
• 认识论，4，11，148，155，157，175，177，188
• Epistemology, 4, 11, 148, 155, 157, 175, 177, 188
• 实验人文学科，192–196
• Experimental humanities, 192–196
• Exploitationware，115–116
• Exploitationware, 115–116
• 扩展思维，37，40，43
• Extended mind, 37, 40, 43

• Facebook
  • 算法套利和111
  • API 的，113
  • 应用程序，113–115，149
  • 阿拉伯之春和，111，186
  • 套利和，111
  • 行为效应，以及，113
  • 黑匣子，169
  • 商业模式，20
  • Cow Clicker和，12，116–123
  • 文化加工和 178
  • 登顿，170
  • FarmVille和 114–116, 119, 122, 124, 126
  • 游戏化，114–115
  • 淘金热，113
  • 排水沟问题，110
  • 拥抱和，113，149
  • 记者和，170，172
  • 点赞数：107，178
  • 媒体公司和 170–173
  • 元数据和 177
  • 挖掘价值，178
  • 使命宣言，9
  • Netflix 集成以及 91、110
  • 作为新闻编辑，170，210n35
  • 布川和，53
  • 可编程文化，169
  • 公共治理和 111
  • 社会算法，20
  • 使用人口统计数据，120
• Facebook
  • algorithmic arbitrage and, 111
  • API of, 113
  • apps for, 113–115, 149
  • Arab Spring and, 111, 186
  • arbitrage and, 111
  • behavioral effects and, 113
  • black box of, 169
  • business model of, 20
  • Cow Clicker and, 12, 116–123
  • cultural processing and, 178
  • Denton on, 170
  • FarmVille and, 114–116, 119, 122, 124, 126
  • gamification and, 114–115
  • gold rush of, 113
  • gutter problem and, 110
  • hugs and, 113, 149
  • journalists and, 170, 172
  • likes on, 107, 178
  • media companies and, 170–173
  • metadata and, 177
  • mining for value and, 178
  • mission statement of, 9
  • Netflix integration and, 91, 110
  • as news editor, 170, 210n35
  • Nunokawa and, 53
  • programmable culture and, 169
  • public governance and, 111
  • social algorithm of, 20
  • usage demographics of, 120
• 信仰，7–9, 12, 16, 78, 80, 152, 162, 166, 168
• Faith, 7–9, 12, 16, 78, 80, 152, 162, 166, 168
• FarmVille（游戏），114–116，119，122，124，126
• FarmVille (game), 114–116, 119, 122, 124, 126
• 联邦通信委员会（FCC），90
• Federal Communications Commission (FCC), 90
• 反馈回路，30–31，60，77，104，112，122，174，178
• Feedback loop, 30–31, 60, 77, 104, 112, 122, 174, 178
• 拜物教，35–36
• Fetishism, 35–36
• 斐波那契数列，17
• Fibonacci sequence, 17
• 《过滤泡泡》（巴黎人），46
• Filter Bubble, The (Pariser), 46
• 大卫·芬奇，98–99，103，106–107
• Fincher, David, 98–99, 103, 106–107
• 闪电小子（刘易斯），12，151，168
• Flash Boys (Lewis), 12, 151, 168
• 福格，BJ，113
• Fogg, B. J., 113
• 富士康，133–134
• Foxconn, 133–134
• 福克斯新闻，170
• Fox News, 170
• 弗雷德金，爱德华，23岁
• Fredkin, Edward, 23
• 从反主流文化到赛博文化（特纳），46
• From Counterculture to Cyberculture (Turner), 46
• 未来生命研究所，191
• Future of Life Institute, 191

• 加洛韦，亚历山大，46，50，54，121–123，143，144
• Galloway, Alexander, 46, 50, 54, 121–123, 143, 144
• 生命游戏，29–30
• Game of life, 29–30
• 游戏化，12，133
  • 成瘾和，114–119，121–122
  • 模糊的现实，120–121
  • 国际象棋和，135–138
  • 文化交易和，119
  • 文化机器和，115–116
  • 深蓝和，135–138
  • 框架和，118–119
  • 开发软件和，115–116
  • Facebook 和 114–115
  • FarmVille和 114–115
  • 信息控制和，122–123
  • 界面经济和，123–131，139–140，145，147
• Gamification, 12, 133
  • addiction and, 114–119, 121–122
  • blurred reality and, 120–121
  • chess and, 135–138
  • cultural transactions and, 119
  • culture machines and, 115–116
  • Deep Blue and, 135–138
  • enframing and, 118–119
  • exploitationware and, 115–116
  • Facebook and, 114–115
  • FarmVille and, 114–115
  • informatic control and, 122–123
  • interface economy and, 123–131, 139–140, 145, 147
• 游戏：算法文化论文集（加洛韦），121
• Gaming: Essays on Algorithmic Culture (Galloway), 121
• 盖茨，比尔，174
• Gates, Bill, 174
• Gawker Media，170–175，210n35
• Gawker Media, 170–175, 210n35
• 性别，60–61，80，82，210n43
• Gender, 60–61, 80, 82, 210n43
• Geocities，209n20
• Geocities, 209n20
• 机器里的幽灵，55，95，183
• Ghost in the machine, 55, 95, 183
• 吉莱斯皮，塔尔顿，20，46
• Gillespie, Tarleton, 20, 46
• 吉列姆，特里，142
• Gilliam, Terry, 142
• 《玻璃笼》（卡尔），38
• Glass Cage, The (Carr), 38
• Gmail，65–66
• Gmail, 65–66
• 哥德尔，库尔特，24，40
• Gödel, Kurt, 24, 40
• 神，1，3–5，7，51，57，71，83，96，113，192
• Gods, 1, 3–5, 7, 51, 57, 71, 83, 96, 113, 192
• 黄金比例，2
• Golden Ratio, 2
• 戈伦比亚，大卫，18，21，38，45–46
• Golumbia, David, 18, 21, 38, 45–46
• 谷歌
  • 广告和，66，74，156，158-160，178
  • 算法套利，111，124，155–156
  • 算法世界观，20
  • 字母表和，66，155
  • 预期和，73–74
  • 作为数字文化的仲裁者，66
  • 增强想象力，186
  • 自动完成数据库和 186
  • 黑匣子，169
  • Brin 和，57，155–156
  • 商业模式，20–21，71–72，93–94，96，155，159
  • 云仓库，131
  • 公司价值，158
  • 文化建筑，42
  • DeepMind 和 28、66、181–182
  • 颠覆性技术，以及 124
  • 收入，158
  • 有效可计算性，以及 42
  • 全球计算基础设施，131
  • Gmail 和 65–66
  • 排水沟问题，110
  • 影响，65–66，87，195
  • 接口和，66–67，124
  • 亲密关系，75–76
  • 知识图谱和，71–73, 75, 94
  • 库兹韦尔和，184
  • 机器学习和，66，181–186，191
  • 地图和，59
  • 市场问题以及 66
  • 大规模基础设施，131
  • Memex 和 188
  • 神经网络和 185
  • 好的，谷歌，51
  • 本体论和，159–160
  • 第155–156页
  • PageRank 和, 20, 111, 155–159, 169, 177–178, 189
  • 解析数据，以及 182
  • 实用主义方法，18，20
  • 产品改进，以及 42
  • 可编程文化，169
  • Project Loon 和 66
  • Schmidt 和，66, 73, 127
  • 搜索和, 26, 42, 48, 69, 75–76, 87, 157–159, 169
  • 共享经济和 127
  • 简化精神，97
  • 星际迷航计算机和，11，65–82，159，186，191
  • 系统行为和 16
  • 技术乌托邦修辞，16
  • X实验室和66
  • YouTube 和 65–66
• Google
  • advertisements and, 66, 74, 156, 158–160, 178
  • algorithmic arbitrage and, 111, 124, 155–156
  • algorithmic worldview of, 20
  • Alphabet and, 66, 155
  • anticipation and, 73–74
  • as arbiter of digital culture, 66
  • augmenting imagination and, 186
  • autocomplete databases and, 186
  • black box of, 169
  • Brin and, 57, 155–156
  • business model of, 20–21, 71–72, 93–94, 96, 155, 159
  • cloud warehouse of, 131
  • company value of, 158
  • cultural architecture of, 42
  • DeepMind and, 28, 66, 181–182
  • disruptive technologies and, 124
  • earnings of, 158
  • effective computability and, 42
  • global computation infrastructure of, 131
  • Gmail and, 65–66
  • gutter problem and, 110
  • impact of, 65–66, 87, 195
  • interfaces and, 66–67, 124
  • intimacy and, 75–76
  • KnowledgeGraph and, 71–73, 75, 94
  • Kurzweil and, 184
  • machine learning and, 66, 181–186, 191
  • Maps and, 59
  • market issues and, 66
  • massive infrastructure of, 131
  • Memex and, 188
  • neural networks and, 185
  • OK Google and, 51
  • ontology and, 159–160
  • Page and, 155–156
  • PageRank and, 20, 111, 155–159, 169, 177–178, 189
  • parsing data and, 182
  • pragmatist approach and, 18, 20
  • product improvement and, 42
  • programmable culture and, 169
  • Project Loon and, 66
  • Schmidt and, 66, 73, 127
  • search and, 26, 42, 48, 69, 75–76, 87, 157–159, 169
  • sharing economy and, 127
  • simplification ethos and, 97
  • Star Trek computer and, 11, 65–82, 159, 186, 191
  • system behavior and, 16
  • techno-utopian rhetoric and, 16
  • X Lab and, 66
  • YouTube and, 65–66
• 谷歌眼镜，66岁
• Google Glass, 66
• 一切谷歌化，68
• Googleization of Everything, 68
• Google Now，51、73–74、76、82、160
• Google Now, 51, 73–74, 76, 82, 160
• 郭台铭，特里，133
• Gou, Terry, 133
• 语法，2，16，25，38–41，62–64，110–112，138，178–179
• Grammar, 2, 16, 25, 38–41, 62–64, 110–112, 138, 178–179
• 侠盗猎车手（游戏），122，124
• Grand Theft Auto (game), 122, 124
• 研磨，120，140
• Grinding, 120, 140
• 《卫报》（报纸），170
• Guardian (newspaper), 170
• 瓜塔里，菲利克斯，76岁
• Guattari, Félix, 76
• 行会，121
• Guilds, 121
• 排水沟问题，110
• Gutter problem, 110

• 于尔根·哈贝马斯, 105–107, 109–110, 114, 172–173, 175–176
• Habermas, Jürgen, 105–107, 109–110, 114, 172–173, 175–176
• 黑客，1–5，38，46，50–51
• Hackers, 1–5, 38, 46, 50–51
• 黑客（电影），3
• Hackers (film), 3
• HAL计算机，181
• HAL computer, 181
• 《事实的半衰期》（Arbesman），188–189
• Half-Life of Facts, The (Arbesman), 188–189
• 停止状态，41–46
• Halting states, 41–46
• 哈特，迈克尔，145
• Hardt, Michael, 145
• 黑斯廷斯，里德，97–98
• Hastings, Reed, 97–98
• 海尔斯，N.凯瑟琳，21–23、28–31、39–40、54、76、93、191
• Hayles, N. Katherine, 21–23, 28–31, 39–40, 54, 76, 93, 191
• HBO，98
• HBO, 98
• 海德格尔，马丁，53，118
• Heidegger, Martin, 53, 118
• 她（电影）, 11, 77–85, 154, 181, 185
• Her (film), 11, 77–85, 154, 181, 185
• 高频交易（HFT），151–158，168–169，177
• High frequency trading (HFT), 151–158, 168–169, 177
• 希尔伯特·大卫，23岁
• Hilbert, David, 23
• 印度-阿拉伯数字，17
• Hindu-Arabic numerals, 17
• 《银河系漫游指南》（亚当斯），123
• Hitchhiker’s Guide to the Galaxy, The (Adams), 123
• 稳态调节器，199n42
• Homeostat, 199n42
• 纸牌屋（电视剧）
  • 美学，92，98–112
  • 芬奇等，98–99，103，106–107
  • 第四面墙，106–107
  • 框架，103–109
  • Netflix 和 11、54、92、98–112
  • 个性化和，98–103，109
  • 斯派西等，98–99，106–107
• House of Cards (TV show)
  • aesthetics of, 92, 98–112
  • Fincher and, 98–99, 103, 106–107
  • fourth wall and, 106–107
  • framing, 103–109
  • Netflix and, 11, 54, 92, 98–112
  • personalization and, 98–103, 109
  • Spacey and, 98–99, 106–107
• 《我们如何成为后人类》（Hayles），28，30
• How We Became Posthuman (Hayles), 28, 30
• 人类智能任务（HIT），135，139，141，145
• Human intelligence tasks (HITs), 135, 139, 141, 145
• 《人之为人》（维纳），27–28
• Human Use of Human Beings, The (Wiener), 27–28

• IBM，135–138
• IBM, 135–138
• 意识形态，130，194
  • 算法，7、9、18、20–23、26、33、38、42、46–47、54
  • 人工智能（AI）和 64、69
  • 比特币和 161–162、167、169
  • 计算主义方法，42
  • 谷歌搜索和 160
  • 高频交易（HFT）和155
  • 接口和 130
  • 劳动和，144
  • 本体论和68
  • 可编程培养和，169–175
• Ideology, 130, 194
  • algorithmic, 7, 9, 18, 20–23, 26, 33, 38, 42, 46–47, 54
  • artificial intelligence (AI) and, 64, 69
  • Bitcoin and, 161–162, 167, 169
  • computationalist approach and, 42
  • Google search and, 160
  • high frequency trading (HFT) and, 155
  • interfaces and, 130
  • labor and, 144
  • ontology and, 68
  • programmable culture and, 169–175
• IEEE计算机协会，6
  • 想像力
  • 抽象和，185，189，192，194
  • 算法，11，55–56，181–196
  • 预期和，73，79
  • 增强，186–189
  • 认知和，182，185，188–189，191–193
  • 计算主义方法，183–185, 192
  • 创造力和，46，56
  • 文化，13
  • 文化机器和，55，147
  • 欲望和，189–192
  • 不同模式，55
  • 有效可计算性，46，93，192–193
  • 爱因斯坦，181
  • 富有同情心，147
  • 启蒙运动和，184
  • 实验人文学科，192–196
  • 智力和，181–183，186
  • 接口和 189
  • 语言和，38–39，185，196
  • 机器学习和，181–186
  • 意义和，184
  • 隐喻和，183–184，189
  • 本体论和，69，73–74
  • 共享经济，148
  • 软件和，186，194
  • 道德行为，以及 146
• IEEE Computer Society, 6
  • Imagination
  • abstraction and, 185, 189, 192, 194
  • algorithmic, 11, 55–56, 181–196
  • anticipation and, 73, 79
  • augmenting, 186–189
  • cognition and, 182, 185, 188–189, 191–193
  • computationalist approach and, 183–185, 192
  • creativity and, 46, 56
  • cultural, 13
  • culture machines and, 55, 147
  • desire and, 189–192
  • different modes of, 55
  • effective computability and, 46, 93, 192–193
  • Einstein on, 181
  • empathetic, 147
  • Enlightenment and, 184
  • experimental humanities and, 192–196
  • intelligence and, 181–183, 186
  • interfaces and, 189
  • language and, 38–39, 185, 196
  • machine learning and, 181–186
  • meaning and, 184
  • metaphor and, 183–184, 189
  • ontology and, 69, 73–74
  • sharing economy and, 148
  • software and, 186, 194
  • virtuous action and, 146
• IMDb，96
• IMDb, 96
• 实施，47–52
• Implementation, 47–52
• 不完整性，24，40
• Incompleteness, 24, 40
• 信息论，10，27
• Information theory, 10, 27
• 知识产权，93
• Intellectual property, 93
• 智能助理，11、57、62、64-65、77。另请参阅Siri；星际迷航计算机
• Intelligent assistants, 11, 57, 62, 64–65, 77. See also Siri; Star Trek computer
• 界面经济
  • Airbnb 和 124
  • 亚马逊和124
  • 套利和，123–131，139–140，145，147
  • 类和，129–130
  • 云仓库和 131–145
  • 高效获取，127–128
  • 假装真诚，146–147
  • 谷歌和124
  • 个人主义，126–127
  • 亲密关系，129
  • 劳动和，123–145，147
  • Mechanical Turk 和 135–145
  • 道德机制和，144–149
  • Netflix 和 124
  • 资本主义的再动员，127
  • 共享经济和，54，123，127-129，145，148
  • Uber 和 123–133、145、147
  • 工人条件和，132–134，139–140
  • 算法的工作，123–145
  • 包装纸和 129
• Interface economy
  • Airbnb and, 124
  • Amazon and, 124
  • arbitrage and, 123–131, 139–140, 145, 147
  • class and, 129–130
  • cloud warehouses and, 131–145
  • efficient access and, 127–128
  • faking sincerity and, 146–147
  • Google and, 124
  • individualism and, 126–127
  • intimacy and, 129
  • labor and, 123–145, 147
  • Mechanical Turk and, 135–145
  • moral machinery and, 144–149
  • Netflix and, 124
  • remobilization of capitalism and, 127
  • sharing economy and, 54, 123, 127–129, 145, 148
  • Uber and, 123–133, 145, 147
  • worker conditions and, 132–134, 139–140
  • work of algorithms and, 123–145
  • wrappers and, 129
• 界面效应（加洛韦），143–144
• Interface Effect, The (Galloway), 143–144
• 接口
  • 抽象和，52，54，92，96，103，108，110–111
  • API，7，113
  • Bogost 和 49
  • 清洁，8，96，110
  • 社区和 52
  • 有意识的，36
  • 文化加工和 16
  • 定制，36
  • 恋物癖和35
  • 谷歌和，66–67
  • 想象力和，189
  • 层数，12，52，123，126–131，140–141，144，189
  • 隐喻和，25，60
  • 拉姆齐和，52
  • Siri 和 59–60、63、75、77
  • 星际迷航计算机和，67–68
  • 透明度和 189
  • Uber 和 54
  • 视觉抽象，25
• Interfaces
  • abstraction and, 52, 54, 92, 96, 103, 108, 110–111
  • API, 7, 113
  • Bogost and, 49
  • clean, 8, 96, 110
  • community and, 52
  • conscious, 36
  • cultural processing and, 16
  • customized, 36
  • fetish and, 35
  • Google and, 66–67
  • imagination and, 189
  • layers of, 12, 52, 123, 126–131, 140–141, 144, 189
  • metaphor and, 25, 60
  • Ramsey and, 52
  • Siri and, 59–60, 63, 75, 77
  • Star Trek computer and, 67–68
  • transparency and, 189
  • Uber and, 54
  • visual abstractions and, 25
• 亲密关系
  • 算法和，4，11，35，54，65，74–78，82–85，97，102，107，128–130，172，176，185–189
  • 谷歌和，75–76
  • 界面经济和129
  • 意义和，75
  • Memex 和，186–189, 195
  • 采矿价值和，176–177
  • 萨曼莎（她）和，77–85，154，181
• Intimacy
  • algorithms and, 4, 11, 35, 54, 65, 74–78, 82–85, 97, 102, 107, 128–130, 172, 176, 185–189
  • Google and, 75–76
  • interface economy and, 129
  • meaning and, 75
  • Memex and, 186–189, 195
  • mining value and, 176–177
  • Samantha (Her) and, 77–85, 154, 181
• iTunes，161
• iTunes, 161

• 杰克逊，塞缪尔·L，59岁
• Jackson, Samuel L., 59
• 詹金斯，亨利，102
• Jenkins, Henry, 102
• 斯嘉丽·约翰逊，78岁
• Johansson, Scarlett, 78
• 琼斯，斯派克，11，77–79，84–85
• Jonze, Spike, 11, 77–79, 84–85
• 记者，3
  • 自动化和38
  • 比特币和12
  • 文化价值观和，171–172
  • Facebook 和 116, 170, 172
  • 游戏化，116
  • Gawker Media 和，170–175，210n35
  • 谷歌和 75
  • Siri 和 58
  • Thiel 和，170–171
  • 交易算法和 151
  • “热门话题”小部件和 180
  • Uber 和 129
• Journalists, 3
  • automatization and, 38
  • Bitcoin and, 12
  • cultural values and, 171–172
  • Facebook and, 116, 170, 172
  • gamification and, 116
  • Gawker Media and, 170–175, 210n35
  • Google and, 75
  • Siri and, 58
  • Thiel and, 170–171
  • transactional algorithms and, 151
  • “Trending Topics” widget and, 180
  • Uber and, 129

• 凯尔·波琳（Kael, Pauline）175
• Kael, Pauline, 175
• 卡斯帕罗夫，加里，135–138
• Kasparov, Gary, 135–138
• Kindle，195
• Kindle, 195
• 基尔申鲍姆，马太福音，47–48
• Kirschenbaum, Matthew, 47–48
• Kiva 系统，134
• Kiva Systems, 134
• 克莱恩，罗纳德，31岁
• Kline, Ronald, 31
• 知识图谱，71–73、75、94
• KnowledgeGraph, 71–73, 75, 94
• 高德纳，唐纳德，17–18
• Knuth, Donald, 17–18
• 库兹韦尔，雷，184
• Kurzweil, Ray, 184

• 工党，7，18，46，122
  • 亚当·斯密，146
  • 情感，145–148
  • 套利和，97，112，123-145
  • 比特币和，164，178
  • 资本主义和，165
  • 云仓库和 131–445
  • 文化机器和，93，119
  • 深层结构，123
  • 假装真诚，146–147
  • 反馈系统，145–148
  • HIT 和 135, 139, 141, 145
  • 身份和，146–147
  • 知识分子，12
  • 界面经济和，123–145
  • 游戏，120
  • 强制性微笑，146
  • 马克思，165
  • Mechanical Turk 和 135–145
  • 采摘者和，132–134
  • 泰勒主义和93
  • 工人条件和，8，132–134，139–140
• Labor, 7, 18, 46, 122
  • Adam Smith on, 146
  • affective, 145–148
  • arbitrage and, 97, 112, 123–145
  • Bitcoin and, 164, 178
  • capitalism and, 165
  • cloud warehouses and, 131–445
  • culture machines and, 93, 119
  • deep structures of, 123
  • faking sincerity and, 146–147
  • feedback systems and, 145–148
  • HITs and, 135, 139, 141, 145
  • identity and, 146–147
  • intellectual, 12
  • interface economy and, 123–145
  • ludic, 120
  • mandatory smiles and, 146
  • Marx on, 165
  • Mechanical Turk and, 135–145
  • pickers and, 132–134
  • Taylorism and, 93
  • worker conditions and, 8, 132–134, 139–140
• Lambda演算，24
• Lambda calculus, 24
• 朗格卢瓦，加纳埃莱，111
• Langlois, Ganaele, 111
• 语言
  • 抽象和，2，24
  • 广告和 178
  • 算法和，24–28，33–41，44，51，54–55
  • 认知和39
  • 颜色词和 4
  • 文化机器和39–40
  • 认识论层次和 4, 11, 148, 155, 157, 175, 177, 188
  • 信息精神，159
  • 语法和，2，16，25，38–41，62–64，110–112，138，178–179
  • 想象力和，38，185，196
  • 不完整性，24，40
  • 作为智力技术，4
  • 智能助理，11、57、62、64–65、77
  • 机器学习和 2, 112
  • 许多寄存器，1-2
  • 数学和，2，55
  • 意义和，1
  • 隐喻和，183–184（另见隐喻）
  • 自然语言处理（NLP）以及 62–63
  • 新媒体，112，122
  • 可塑性与，38，191
  • 幂，1–2，4–5
  • 程序性，3–4，6
  • 现实和，1
  • 修辞和，6，16，22，30，45，89，96，101，104，110，112，123，127，136
  • Siri 和 57–65、71–84
  • 口语，2，58，60，62–63，67，84，185
  • 象征性的，2，26，38–41
  • 技巧和，3-4
  • 图灵机和，33，41
  • 通用，5
  • 词汇和, 2, 4, 25, 138, 160, 190
  • 维纳和，28
• Language
  • abstraction and, 2, 24
  • advertisements and, 178
  • algorithms and, 24–28, 33–41, 44, 51, 54–55
  • cognition and, 39
  • color words and, 4
  • culture machines and, 39–40
  • epistemological layers and, 4, 11, 148, 155, 157, 175, 177, 188
  • ethos of information and, 159
  • grammar and, 2, 16, 25, 38–41, 62–64, 110–112, 138, 178–179
  • imagination and, 38, 185, 196
  • incompleteness and, 24, 40
  • as intellectual technology, 4
  • intelligent assistants and, 11, 57, 62, 64–65, 77
  • machine learning and, 2, 112
  • many registers of, 1–2
  • mathematics and, 2, 55
  • meaning and, 1
  • metaphor and, 183–184 (see also Metaphor)
  • natural language processing (NLP) and, 62–63
  • of new media, 112, 122
  • plasticity and, 38, 191
  • power of, 1–2, 4–5
  • procedural, 3–4, 6
  • reality and, 1
  • rhetoric and, 6, 16, 22, 30, 45, 89, 96, 101, 104, 110, 112, 123, 127, 136
  • Siri and, 57–65, 71–84
  • spoken, 2, 58, 60, 62–63, 67, 84, 185
  • symbolic, 2, 26, 38–41
  • tricks and, 3–4
  • Turing Machine and, 33, 41
  • universal, 5
  • vocabulary and, 2, 4, 25, 138, 160, 190
  • Wiener and, 28
• 新媒体语言（Manovich），122
• Language of New Media, The (Manovich), 122
• 诉讼，90，171，175
• Lawsuits, 90, 171, 175
• 莱布尼茨，戈特弗里德·威廉，25–27，72
• Leibniz, Gottfried Wilhelm, 25–27, 72
• 莱姆，斯坦尼斯拉夫，184
• Lem, Stanislaw, 184
• 利维，史蒂文，3
• Levy, Steven, 3
• 刘易斯，迈克尔，12，151，153，168
• Lewis, Michael, 12, 151, 153, 168
• 莱顿，彼得，160
• Leyden, Peter, 160
• 图书馆计算机访问/检索系统（LCARS），67–68
• Library Computer Access/Retrieval System (LCARS), 67–68
• 《生活》杂志，31
• Life magazine, 31
• 识字, 5, 39, 52, 75, 109, 129, 159, 177
• Literacy, 5, 39, 52, 75, 109, 129, 159, 177
• LiveJournal，209n20
• LiveJournal, 209n20
• 洛布纳奖，87，203n50
• Loebner Prize, 87, 203n50
• 逻辑
  • 普遍替代性，以及 33
  • 哥德尔和，24，40
  • 停止状态，41–46
  • 信息理论和，10，27
  • 无形的排斥，110
  • 实用主义方法，18–25，42，58，62
  • 过程和，41–46
  • 证明和，15，24–25，41，44
  • 理性，38，40
  • 象征性的，2、21、24、39、41、44、54–55
• Logic
  • general substitutability and, 33
  • Gödel and, 24, 40
  • halting states and, 41–46
  • information theory and, 10, 27
  • invisibly exclusionary, 110
  • pragmatist approach and, 18–25, 42, 58, 62
  • process and, 41–46
  • proofs and, 15, 24–25, 41, 44
  • rationality and, 38, 40
  • symbolic, 2, 21, 24, 39, 41, 44, 54–55
• 《长臂》（Schwartz and Leyden），160–161
• “Long Boom, The” (Schwartz and Leyden), 160–161
• Lyft，123，127-130，145，148
• Lyft, 123, 127–130, 145, 148